[ortek_service]

Thanks for posting your findings on this.
I presume putting x10 etc. 'scope probes on the input and output, would show how long the FET is taking to turn off relative to input signal, to gauge how much more power-saving should be possible with a more suitable FET / better drive to it.

MOSFET input capacitance is generally proportional to output current (so inversely proportional to RDSon). So rather than using a 4A one, if a much lower rated one is used then input capacitance wll be much lower.
e.g. Just checking the spec. of a popular P-Ch FET, the BSS84 (Id(Max)=130mA), this has an input capacitance of 25pF typ.

So this should do for a single PROM, as you've used one for each, but would need something a bit better higher Id(Max) rating to power both PROM's from a single device (but maybe not as easy to implement with single switch on the MK14's layout).
And a parametric search on Digikey etc. could help find some more optimum ones.
Driving the MOSFET with a fairly low Rds(on) N+P Ch. CMOS style output stage, would also help to turn it off very quickly, compared to open-drain resistor pull-up RTL style logic.

So don't necessarily have to resort to a bipolar PNP device (But maybe could try something like a BC212 (Ic=200mA Max.) or BC327 (Ic=500mA Max) instead.