That's interesting. As far as your video RAM is concerned.... have you tested the 2114s? I've seen many non-working (possibly fake) 2114s over the last year or so... It might be worth randomly picking a pair of 6550s and using those. The fact we have seen rendered characters suggests the character ROM is working... Did you prove the old one wasn't?
I'm tempted to guess that the CPU is working with the ROM/RAM board... just not sure how we prove it...
To write to the video RAM requires the CPU to get through two sets of buffers...
View attachment 1250076
The G5/G6 buffers and the B3/B4 buffers and at the same time make sure the H8/H9 buffers aren't interfering.
The next step is easy with a NOP generator as the address bus changes in a known way, however with a program running there should be some activity (but perhaps not across all the bits).
You could replace the ROM/RAM board and then look at E5, E4, E3 and E2 which are used for buffering the address bus.
View attachment 1250078
Obviously all bits matter but BA11 might be a good place to start. Put your logic probe over E5 (and E4 and E3 and E2) and look for stuck bits.
Note: E5 is a set of inverters, E4 and E3 are non-inverting buffers, E2 is a set of AND gates. That's why BA11 goes through 2 gates... NOT NOT... obviously they had some spare gates at E5.