The only drawback is that the chip needs to be programmed with this code at least once. This is no problem for someone with a parallel programmer. For me, I guess I will require some jumpers to bootstrap it into that condition.
Where is the drawback, if you did not use NoTouch, you would have to do it every time.

(P2.7, 2.6 would always be pulled high on a reset, ALE always disconnected).
P2 pins 6 and 7 are ok as is if no output from elsewhere ia attcahed I would have the 2 10k resistors there just to feel warm and fuzzy.

Ah dunno, what is DTR connected to on the RD2
FlashMagic allow a watchdog reset at end of ISP.

What I was thinking was to put a jumper on P3.7 (an available pin), to pull it to ground (is it safe to go directly, or should I use a 2k resistor?) if I want to go into programming mode. This pin would be continually checked in my main while loop.
I like TxD, It is always high when not in use, has a weak pullup and is used, thus no pin is lost.

So now I'm wondering about your assembly code that you have provided in the NoTouch document.
See Philips appnote AN416

Shouldn't the watchdog timer be enabled at some point?
That is always the case for me

Erik