Slothie, SH has been singing your praises - credit for making your own PCB by the way...
Looking at the signal on C3 on my JM there is quite a bit of noise on the 5V measured while using my NorthPada 9V SM PSU - which I use for my Spectrum Next and my DSO small scope.
Hopefully I have remembered how to use the bigger scope to show the pulsed nature... feel free to point out the errors in my settings it has been a while since I did this sort of thing.
Take any measurement accuracy with a pinch of salt: XIN shows a clock pulse on 0.5uS per div with just under four peaks across two divisions so probably 3.8MHz so a bit low for the 4MHz crystal - it's not really a square wave but a sine wave which I assume is due to the limits of the scope - I was sure it was 20MHz but, maybe I was wrong.
I will change over the connector polarity tomorrow and try a Spectrum PSU to see if that is gone and has an effect on the reset.