One question: is it possible to increase the maximum program memory (e.g., to 64KiB)? 256 opcodes are far too few for my current program. I would then have to outsource it to over 30 fCPU´s and always copy the current data between them to the master CPU.
Or would that cause performance problems?

It would be really cool if that were possible!

The main limitation is the blueprint size. There is no other reason for the 256-instruction limit.
I think the instruction count could be seamlessly increased to 1024 or so.
Let me check something and I'll reply in a few days!