I contend that the low-hanging fruits are largely about expanding the proportion of the metro population that lives in the city proper, and not about expanding the population of the metro area. At the size class of New York, the limits to further growth are diseconomies of scale, coming from commute length (Tokyo, which is 50% larger, has even worse commutes, averaging about an hour).
The exception is that nowadays, there's a secondary limit, coming from zoning restrictions. If it were legal to build higher in New York, there would be more construction in city neighborhoods near the subway, and more office construction in hot areas outside Midtown, for example the Meatpacking District.
So really you should ask what kind of post-1783 changes would allow New York to capture more of its suburbs, and butterfly away restrictive zoning. As it happens, there is a model of a city that did well on both accounts: Toronto, which annexed many of its suburbs in the 1990s, and which (like all major cities in Canada) has a lax zoning regime by American and European standards.
Getting to Canadian zoning is not hard with an 18c POD. There is a very strong causation arrow running from deciding zoning at a high level of government to pro-growth policies. In Canada, approving development is a provincial responsibility, and in Japan, which has even laxer zoning rules, it is decided at the national level. Within the US and Europe, zoning is decided at the municipal level, where NIMBYs have more power. What's more, large cities, where the municipal level is more removed from the level of the neighborhood NIMBY, are more pro-growth than small ones. For a good natural experiment, compare San Jose with Silicon Valley edge cities like Palo Alto and Mountain View. San Jose and the Peninsula cities share the same approximate political culture and regulatory regime, and are both thoroughly suburban in form, but San Jose happens to have had loose municipal boundaries and is now a city of a million people, and Palo Alto and such grew as small suburbs, and have municipal government that are closer to the NIMBYs. San Jose approves development from time to time, and maintains a healthy growth rate, whereas Palo Alto and such are NIMBYvilles, permitting almost no residential growth and extorting benefits from tech companies that want to add jobs.
Bottom line: if you want a bigger New York, you want some late-19c POD under which zoning rules would be a state responsibility rather than a local one. You can even get away with a mid-20c POD - New York is a huge city, but tends to defer to local community boards when it comes to any development that's not backed by multibillion-dollar slush funds. Columbia can expand over the community board's wishes, but some random developer who wants to replace a 6-unit building with a 20-unit building has no chance.
Now, let's discuss putting more of the metro area within the city proper. Obviously, there's no hope for expansion across state lines - it's the 18c, not the 17c. But it's possible to have most of Westchester and western Long Island's population remain in the city proper. This requires finding ways to prevent the American pattern in which suburbanization began with the rich, and continued with the middle class. Brookline resisted annexation in the late 19c because it was already rich, and postwar suburbia resisted having anything to do with the city because it was middle-class and white and the city was poor and increasingly black.
The flip side is that if the suburbs are poor and the city is rich, the city does not want to annex. Paris is tiny - 2.5 million in a region of 12 million - because the city resisted annexing the working-class suburbs, especially in the postwar era, when the honchos were afraid the suburban workers would elect a communist mayor. But note also that metro fragmentation can happen in environments with relatively even incomes, such as Vancouver; relatively even incomes are a necessary but not sufficient condition.
If you want a specific POD that would have the highest chance of disrupting the American pattern of suburbanization, have urban renewal succeed earlier. Let's say that New York bashed heads together and opened the subway in 1870 and not 1904, and cleared the slums in 1890 already. The cleared poor would live in terrible project-like neighborhoods in Upper Manhattan, and the whole process would be obviously a flop by the 1910s and 20s. There would be more successful middle-class projects, a whole bunch of Radburns, but overall there would be a lot of poverty in Upper Manhattan. This looks like OTL, but the key difference is that this predates the Great Migration. In the 1920s, the rich are happy to stay on the Upper East Side and don't begin the push for suburbanization; the suburbs still grow, but they have the demographics of Yonkers and New Rochelle, and not those of Scarsdale. By the 1950s, there's no anti-city mentality among the elites, so there's less resistance to having the city take over most or all of Westchester and Nassau Counties.
With the urban renewal failure happening in 1890-1910 and not 1950-1970, restrictive zoning never happens. In OTL, the 1961 zoning code was a reaction to Robert Moses's projects. A similar reaction in 1901 would have been unlikely, given massive open land for urban growth that did not require replacing buildings. CBD growth was replacing buildings, but not in any neighborhood people cared about.
The bottom line is that by 2016, the New York metro area has 23 million people, as in OTL, but without New Haven and Allentown. But the city proper has 13.5 million people - about 11 million in constant geography, plus 2.5 million in annexed Westchester and Nassau neighborhoods (the sixth and seventh boroughs). Jersey City is a consolidated city-county with 900,000 people, and Newark is another consolidated city-county with 1.3 million people. Suffolk County only has 1 million people (OTL 1.5 million), the Bergen-Passaic area has about 1.7 million people (OTL 1.4 million), points north about 700,000 (OTL 1.3 million), suburbs in Connecticut 700,000 (OTL 900,000 in constant geography, 2 million including New Haven), and suburbs in Central Jersey and the Jersey Shore 3.2 million (OTL 3.6 million). Without an anti-urban bias among the elite, there's more investment into public transit in the 1950s and 60s and less in highways; total ridership on all rail systems in the region is 3.5 billion (OTL 2.1 billion).