This particular cap is a ceramic, 0.1uF -- Tied from pin 3 of G5 (Reset*) to ground (G5 pin 7). It is used to stretch the length of the reset signal.
These don't leak and seldom fail. But if it does go open internally, it can result in a shorter reset signal.
Can't measure with regular meter but can check with cap meter - disconnect to check. But since these caps are so cheap, if you're going to disconnect it then you may as well replace. Also, G5 is a CMOS type integrated circuit which means it can easily fail due to ESD if mishandled.

If you replaced the CPU - what did you use? There is only one brand that is still made (WDC), the rest are old stock from many years ago. A ton of the "Rockwell" marked parts are from China as reclaimed counterfeits.

Post #8 above shows a bunch of schematics from this board. Scroll to sheet 44. The cap you show is at top, center of his sheet 44. Indicated as "Located on foil side" in schematics.

The cap is possible but unlikely.

There is no power-up reset on this board - it relies on the reset being sourced by the CPU board.
Since manual reset works - this says G5 works (or maybe just input 1 bad but unlikely).
Follow the reset line from the sound board back to CPU board. Probably a bad connection between the two.

Added to post #23. But just to be clear -- it doesn't work until you reset it manually and then it works?

Then, my first guess would be a bad connection between the two boards.