1) Thanks for catching the typo. I've edited the earlier post to indicate WNYW. So your list of NYC priority stations by call sign and real channel number are:
WNYW - 44
WNBC - 28
WCBS - 33
WABC-7
and from Philadelphia, WXTF - 42. I presume anything else is desirable, just not at the top of the list.
2) The HD8200 is going to do well with real UHF channels listed on your report down to the bottom of the section with red background. As the NM values turn negative, it becomes more likely that you'll have trouble with reliable reception. As I look at WNYW with a predicted NM of -10.6 I add the gain of the HD8200, about 12 dB, I get a net NM of +1 to +2 dB that can be delivered to the preamp. The noise figure of the preamp is about 2 dB, which subtracts from the net NM. So you'd be left with a net NM at the preamp output of 0 plus or minus a dB or two. If your net NM at the preamp output was 10 dB or higher, you'd hope to have fairly reliable reception, a NM of 0 dB is the lowest theoretical value that an excellent tuner will lock on... but there is no provision for noise, interference or fading. The larger UHF only antennas give you a little margin but it's not going to be 10 or more dB as I would like to aim for.
Even though there is more noise to contend with in the low VHF band, the HD8200 should see WPVI. I don't expect you to see WPXO or WKOB though.
3) Avoid cheap noisy amplifiers, you don't have any noise margin to waste.
4) Presuming you use 1/4 wave length transmission line sections to do the combining and matching, stacking/ganging for gain will make a broadband antenna into a somewhat tuned, narrow band antenna.
stacking/ganging to null adjacent and co-channel interference is easier to do without as much narrowing of the useful bandwidth of the antenna if you use a reversed hybrid splitter. The loss of the splitter will leave you with slightly less net gain than one lone antenna.
Bottom line, stacking/ganging an all channel antenna would be a waste of a whole lot of bandwidth in the majority of cases.