The issue seems to be with your system. The problem should be with your PCB. Some how the internal pull up of your XRES has been fused off. You can check with an external pull up (do not short). I am also thinking that the exploding FETs and all are because of this, as you deduced. If you could share your schematic, we can check what is the issue.
I think that the XRES is being driven. This is because I can use a 1k pull up, and it makes no difference. The impedance to GND is quite low... can't think of another reason except that it's being driven by a source. The absense of a pull up would just give a floating pin right?
Or is it possible it's fusing to gnd... ? Would a negative voltage on that pin latch it down?
I won't bother putting sch up, found the issue. Looked back through notes found that there was a problem with a dual bat54 package burning out. It was a solder fault that connected 24v to either gnd or 3v3 under the package and the trace blew off.
This has eigher negative biased the PSOC by connecting gnd to 24V or just over voltaged it 3.3V becoming 24V until the fuse current burnt the short out. The device held on for a little while then the XRES let go... It also seems that it has only happened on one board reused in two systems, hence my though it was two chips failing.
I'll replace it and see how things go.