For a limited selection of opcodes, we can allow updates through OSC.
This implements an "opcode buffer" that gets read at the beginning of the render loop in the RT thread.
Any opcode updates must thus be RT-safe, so not all of them should be allowed.
I added a parameter to disable the opcode name cleanup method which does allocate strings potentially.
This means that the opcodes must be "correctly" normalized.
For example, for `lfoN_wave`, we have to pass `lfoN_wave1` in the opcode buffer.
