If you add up all the people whose commute crosses New York - for example, from anywhere in Jersey to Brooklyn, Queens, and Long Island - you get about 200,000 people, i.e. 400,000 weekday trips. The reason those infill stations (and the extra connections, and the integrated fares and schedules) are important is that they make it easier to take rail to secondary job centers. To Manhattan practically everyone rides trains anyway; the only major exception is the GWB commute shed, and that's a problem of the Erie Lines not serving Manhattan directly and Secaucus being an awful transfer with multiple level changes and internal faregates.
Commuter volumes also drop precipitously as distance grows. So of the above 200,000 people, if you e.g. look at those who work in Queens, there are 30,000, of whom 17,000 live in Westchester, Staten Island, Bergen County, and Hudson County, and another 5,000 live in Essex, Rockland, Middlesex, and Monmouth Counties. MOM and other outbound extensions are useful to a small subset of these; being able to make the commute on one train, or two trains with a relatively tame connection, is much more useful. The West Shore Line and the Northern Branch both serve relatively strong suburban markets, to both Manhattan and the job centers to its south and east, but the other possible extensions aren't as important. Few people live along the West Trenton Line, for instance.
Now,
1. There are no plans for a fare union between urban transit and commuter rail on the New York side, or for one between NJT and the MTA agencies.
Yeah, I know not everyone thinks it's necessary. But unified, mode-neutral fares is how European cities have done their transit revival. (North American ones, too - all bus, ferry, and SkyTrain fares are integrated in Greater Vancouver; the commuter rail fares are separate, but they're a tiny portion of systemwide ridership.)
2. That's on the branches, not in the main trunk tunnels. NJT was planning and Amtrak is planning a tunnel to relieve the North River Tunnels, with no plans to increase throughput on the existing tunnels using better signaling.
3. NJT is still buying locomotives. It's even buying dual-mode locos, at much higher cost and weight than is normal, because electrifying the Erie Lines is for wimps, or something. Really, the only lines that have any business staying diesel in 2050 are Harlem north of Southeast and Waterbury, both of which have approximately zero riders and not enough of a commute market to dense or densifiable CBDs (including even secondary ones like White Plains) to matter
4. Are there concrete funded plans, or wishlists similar to Amtrak's 25 kV wishlists from the 1970s and 80s?
5. TOD is hit or miss. Stamford has some plans, which the local commuters are up at arms about because they think they're entitled to a parking garage right at the station. Hicksville doesn't have any that I know of - instead it recently rebuilt the garage for about $25,000 per spot, which is about as much as the per-rider cost of Second Avenue Subway. Ronkonkoma doesn't have any, either. The LIRR generally lags, what with the one-way rush-hour operation (which is not required by current traffic, as long as one is willing to give up express trains). But Mineola, which is somewhat better, manages to get a couple hundred more eastbound am disembarkings than Hicksville.
6. Conductors are not necessary for anything except makework. Somehow, the French, the Germans, the Swiss, the Dutch, and every light rail and BRT system in the US manage to do with POP, and the safety record is generally better than on US mainline rail (thanks, Metra). Even Caltrain and Metrolink do POP, but they retain conductors because That's How We've Always Done It. At the staffing levels appropriate for a country with first-world wages, the correct number of employees per train is 1, and 0 for a closed rapid transit system.
7. LIRR trains are stabled in Manhattan west of Penn Station, and NJT trains at Sunnyside Yard, for the midday off-peak. ARC Alt G actually included a Manhattan yard expansion, never mind that it was going to connect Penn Station with Grand Central, which is basically a railyard with a train station annex.
8. Sure, but not every rail expansion is needed. For example, the Northwest Rail Link just splits frequency with more useful lines in the inner parts, and serves lightly populated regions farther out. Not every rail line needs to host commuter rail - you want to make sure the lines have reasonable enough frequency. For example, one of my commenters, Anon256, is already complaining that I'm proposing 5 branches going into an Erie tunnel to Lower Manhattan (Northern, West Shore, Pascack Valley, Bergen County, main), which would cut frequency to those lines to avoid saturating the tunnel.