alternatively you might hang a temporary 10m dipole and test into that. A 10m dipole can be cobbled together in no time as it's just approximately 16' of wire, some kind of center insulator and coax. The antenna doesn't even need to be hung up high in an ideal permanent position, just up off the ground enough to present a reasonable 50 ohm match for testing purposes as that's one of the problem oscillation bands. Now that you've identified and remedied at least a few major issues (extra output network, parasitic suppressors and missing HV bypass cap at the anode choke) have you tried testing with the backup set of tubes you mentioned? They might both have oscillated originally but you might see differing behaviors now that a few things have changed. Also you might double check the T/R relay contacts and burnish them with something like a dollar bill if they're not shiny clean. Adding a 10 ohm 10 to 15 watt, non inductive swamping (aka feed back) resistor between the input tuning networks and the caps that drive signal to the tube cathodes could also help, but really I'm grasping at straws and trying to think of any ways to decrease the loop gain under no-drive conditions that lead to oscillations. I do have a swamping resistor like what's described in my input network but I didn't add it for oscillation reasons but if your feedback loop includes any stray resonances back through the input tuned networks on the higher frequency bands then perhaps that would help. Writing that, it also occurs to me you might try re-tuning the input network on one of the problem bands to see if that does anything to quench the parasitic oscillation. BTW, I'd search around and try to source some carbon comp resistors for those parasitic suppressors as your use of metal film isn't ideal since many film resistors are spiral trimmed for value tuning during manufacturing which can leave them inductive and you're looking for purely resistive loads in those suppressors.