Well I would expect that Boost would make things worse.
What authority do you have that it’s definitely not a VST causing the timing difference?
Your last statement is somewhat ambiguous.
2 midi tracks. One sending. To where? One receiving. From where? Each other?
I suppose the kernel here is: What is this preventing you from doing that countless other users do every day?
Do you need a greater level of accuracy for something?
It seems it happens only during playback and not during exporting the Audio as a mixdown. I exported same midi track twice and played both Audio mixdown together with one of them has flip phase on, and there was no sounds. (Marker 3)
See. This is what I would expect and points to the timing in Cubase to be ok. It also means that the timing inside the VST records to audio constantly. Both are generated by the midi engine (unless you just copied one track) so the midi engine is constant at least as proved by these recordings nulling. (exporting the midi track effectively “plays” it ie: it is not static)
What I would expect in a sample based drum machine is that there will be timing discrepancies due to the sample placements, attack position and FX generated within the VSTi and, like any other synth recording, trying to null against the original midi track will never work as the recording will be fixed while the synth will modulate.