IOC flag triggered when PIC is unplugged...?

[Demon]

That is a false assumption, all the IOC flag says is that a edge was detected, ...

I know that now.

... no simple hw debounce solution will do all the work...

The 74HC14 with Bourns' debounce was doing very well. I'm rewiring that back in now to redo testing.

... for my money i would ignore anything less than 50mS...

I'll post results when I have everything back up and running. It was surprisingly stable before I migrated to a 16F18877.

... switches on long leads and breadboards are potentially difficult cases also

Switches and encoders are all mounted on headers, same for power jack, USB connector and 7805.

The only leads for switches will be from 74HC14 to PIC, and I keep those on 2 breadboard area maximum. The only PIC with long leads is for the LCDs.

The USART leads (in white) will branch out from the USB PIC out to the SW and ENC PICs (the END lead is 5in straight line, so about 6in when strung out).