Yes, that is a perfect summary and interpretation of the issue.
I think my initial suspicion of the issues are accurate, and there is still the compounded issue that FluidNC will insert delay_ms regardless of the current state of M7/8 (and even does it on M9), hence my last post there requesting that it still be included as a feature, since if there was no delay on subsequent M7/8 commands, the issue would also be resolved.
I guess It’s really hard to say who should fix the problem. In some ways I guess it falls back on it being a FluidNC issue because they have implemented a change in behavior from grbl (assuming they are comitted to providing 100% backwards compatibility with grbl), but they are not really obligated to as such.
But there is also a reasonable argument to be made that to be strictly compliant to the grbl ‘standard’, Lightburn should be ensuring that an appropriate command be issued before non G1 commands when in M3 mode to ensure the laser is turned off, not assume the controller will handle it automatically (because officially, grbl does not, only coincidentally).