Yes, construction error indeed - I found that the diode of channel 6 was not connected due to a lifted trace via continuity test, so now I have no more bleed from distant channels.
The problems that are still there are that channel 2 and 4's inputs pass no signal, but accept full bleed from the previous channels (1,3).
Channel 5 distorts the signal, now channels 6, 7 and 8 are mostly muted, with 8 passing signal occasionally and bleeding to 7.
Still no reset.
I know now that there was likely nothing wrong with the 074s. In these last two problematic projects I used .1uf polypropylene caps that I thought would give superior clocking performance, perhaps these were a poor fit.
Thanks,
Eric