You are sticking with FPTP for each small district then?
How do you get these indices of party lean in each district? I suppose you might have access to a database of all the precincts in each state, and either some pundit group has computed these indices based on past electoral behavior plus some sort of demographic evaluation to account for changes since the prior elections, which of course would involve some guesswork as to how the people in the demographics might split. Or you could perhaps have actual 2016 election data, in some huge database precinct by precinct, and if that database includes information on how many people qualified to register to vote, or simply how many registered voters there are, in each district, you could then construct districts by some method.
Unlike actual districts, precincts are not supposed to be equal in population particularly. They are regions formed for the convenience of voters and the electoral tabulation machinery, and quite a few precincts might have just handfuls of people in them while others contain huge numbers of people. As long as the largest precincts contain fewer than 30K people you ought to be able to construct districts by accreting precincts together, but it would be a nightmare I suppose. I speak from the experience of trying to jigsaw existing districts together. If precincts range between practically zero population and some moderate fraction of your 30,000 residents per district target, you'd be constructing all these ten thousand plus districts by hand, using units that have the most bizarre shapes.
So why this precise number by the way? Is it because actually you have stumbled on some pre-sorted districts (Zip Codes? Telephone exchanges?) that already hew closely to 30,000 people per for some systemic reason, and 10273 is what they add up to?
If we could manage a House of Delegates of this size and conduct actual business with such a thing, I would suggest consolidating them by fives into Single Transferable Vote districts, each electing five. That's still 30,000 people per elected delegate but the proportionality of the outcome would be closer than with all five totally separated. Mind, STV has its own shortcomings. But way better than FPTP! I actually would prefer a quite different approach.
So I went ahead and did both.
I have no idea how you arrive at your PVI for each district, but I presume it means percentage above 50 percent for the specified party. As a general thing, American elections even in Presidential elections don't generally involve more than 60 percent of the residents voting--some can't but most of the deficit comes from people who could register to vote, or are registered but don't show up, not voting by choice. Or because they find it tough to make the time, which is quite a hardship for some people in some districts actually.
So I figured a flat 18,000 people voting in each district, and used the PVI you provided to generate a number of votes--if the PVI was +14 R I'd have the spreadsheet write in 18 thousand times (50+14)/100. The two +D districts I handled by treating their pro-D index as a negative Republican index. The Democratic votes would then be the residual from eighteen thousand.
Naturally this makes no attempt to capture any effects third party candidates had.
I then grouped the districts, one using 4 (the southeast, 1-3 plus district 10, that is Cheyenne, the 3rd district surrounding Cheyenne, and neighboring 10) which allowed 3 more groups of 5 each going clockwise, Southwest, Northwest, and Northeast.
Applying Single Transferable Vote rules, I figured I would take the district figures to represent the distribution of first-choice votes for the candidates of the respective parties, with both R and D running 4 in the southeast and 5 in each of the others. We then sum up all ballots cast, and divide by the number of seats plus one to get the Droop Quota--which worked out to be 14,400 ballots in the southeast for 4 seats and 15,000 for the five seat districts. Candidates who exceed the DQ in first choices are marked as having won, and then the DQ is subtracted from each of these, and the remainders distributed evenly to any remaining candidates of their party.
Note that actually STV does not pay attention to party affiliation formally, like FPTP it is in principle a race between individual candidates without reference to party. I just assumed that any D voters in Wyoming would list on their ranked choice ballot all the D candidates, in order of their personal preference for each, before listing any Rs, and vice versa for the Rs. Since I have no way to guess which voters preferring one of the 5 or 4 running for their party would list which alternate in their party next, I just distribute them evenly to the remaining candidates. In real life some voters might mix up the parties, listing say one D first than several R's then another D perhaps. But I suspect that kind of behavior would be rare in this case!
When all quota winners out the gate are elected and their residual votes transferred, elimination begins. Whichever candidate has the lowest first choice votes (including now any redistributed to them by supporters of other candidates who listed them next) is eliminated and all their votes transferred to whomever each ballot lists next. This is like the above bit except we don't subtract the DQ since these candidates are eliminated without being elected; their supporters aren't satisfied with electing anyone yet. This causes the other candidates a voter lists lower down to accumulate votes until one of them pops up above quota.
Doing this with the 4 districts, resulted in one Democrat being elected in each, with 15 instead of 17 Republicans winning the rest. The two eliminated Republicans would be Ed Murray of West Cheyenne, district 1, and Barbara Cubin of District 5, West Casper. Whoever you figured would be the Democrat running there would win instead. Note that this is something of a coincidence; it often happens when I group together OTL districts into larger STV bailiwicks that none of the candidates from some old pre-consolidation districts win and both major party candidates (sometimes a third party candidate too) from another previous district wins; it is all about how many votes they got.
Now when I turn to actual proportionality, I find even this doubling of Wyoming Democrats in the delegation does not fully bring the delegation to proportionality; in fact, taking these numbers of votes inferred as I outlined as real, out of 19 seats, the total statewide Republican vote would entitle that party to 14.21 and Democrats to 4.79 by raw fractions. Applying either Jefferson-d'Hondt or Hamilton's method tells us that to get 19 seats the Republicans would have 14 and Democrats 5--whenever those two methods agree on an exact outcome, one can be sure any method that can call itself "proportional" will give the same answer because these approaches are at opposite extremes, Jefferson's favoring the larger parties more, Hamilton's being most inclusive and favoring smaller parties.
Thus under FPTP the Republicans win 3 more seats out of 19 than the total state vote for them would justify, the Democrats are slashed to 40 percent of their proportioWynal share. The way I would resolve this is to not take any seats away from such an overhanging party, but determine how many seats should be added to make their FPTP gains proportional, then raise up the shortchanged parties by declaring also elected their top also-ran candidates. So here, it turns out if we use Jefferson's method the Republicans are justified in having 17 seats when there are 22 seats altogether; not by accident this is just the number of seats Democrats need to be brought up to proportionality. Therefore, noting Democrats won two seats by majority in the two D+ districts, three more Democrats--in districts 1, 2, and 5 would also be made winners and Wyoming's delegation rises to 22. We've already seen that two of those would win seats under STV, but they'd block two Republicans who won FPTP in your system. Here those three Republicans who won those three districts would be secure in office, but their Democratic rivals would face them in the House.
This system involves counting the votes for candidates in districts three ways--first, it determines who wins the district race, which turns out to be very lopsided.
I presume you didn't gerrymander your districts to make it work that way; it is just in the nature of FPTP voting to be flaky and perverse and deviate far from a proportional outcome. You might think when we go from 20 about seats in a state like Wyoming to 1200 in California, it comes out in the wash then, with numerous districts approximating proportionality, but I am going to predict you will find it otherwise. I do expect a little bit of converging toward overall state PR but only a little.
Then, however the districts turn out, each vote for a candidate counts a second way as a party vote. (We can't meaningfully talk about "proportionality" at all unless we talk about party). This gives us an index to compare the FPTP results to, and thus figure out how much we have to level up the shortchanged parties. Then, having determined how many seats each shortchanged party have earned to be in proportion to the parties that come out ahead, we count the votes a third way, to see which candidates of these upgraded parties ought to be the added representatives--the ones with the most votes have the greatest confidence of voters.
It would also work with STV--after the mechanism operates, we find the R's have won 15 seats instead of 17, and still, taking the first choices of each voter as indication of which party they have most confidence in, the R's are one ahead; we level up to justify their 15 seats, and find this still justifies the D's claim to 5, so we need one extra seat for the D's to make it 20. The highest vote winner, of first choices, for the D's not having already won under STV is again the second district candidate. Now on one hand we have frugally limited the increase to just 5 percent, but on the other, the STV process bumped out two Republicans who won a lot of votes.
The main reason to favor STV in the first place is to allow for an approach to PR without triggering people who think recognizing party in the election mechanism is some kind of abomination, but to level it we have to explicitly recognize that voters often are voting for a party win, and they support these parties for legitimate reasons. Might as well stick with FPTP single member districts and level those then; that way we get the best options for voters who do think community issues are the primary thing.