I think part of the problem is that everyone keeps referring to the M-Drive (or "Thruster Plate") as "Reactionless". To my knowledge, only under MT/DGP:SOM is it explicitly "Reactionless"
1.
1 - DGP:SOM describes the Thruster Plate as being based on the Strong Nuclear Force, and employing some type of interaction that causes the plate to "push-off of itself". This interpretation was not carried forward into successive editions of Traveller.
CT post-1981 is rather ambiguous on the nature of the M-Drive (CT-era DGP considered it a gravitic-based system).
Both T4 & T5
explicitly describe the M-Drive as employing a gravitic-based interaction. Its thrust-efficiency also drops off to 1.00% beyond 1000 diameters under these rulesets.
The reason that is significant is that in any given star system there are plenty of gravitating bodies to react against employing a gravitic-based system, so it is not necessarily a "reactionless" system, any more than a NASA probe utilizing a gravitational slingshot or gravitational breaking maneuver is a reactionless interaction. It is just that the gravitating bodies that the vessel is interacting with are so huge compared to the vessel itself, that the momentum change of the bodies in the star system are unmeasurably small.
So is part of the mechanism of the M-Drive (however it actually works) picking up momentum from other bodies in the system, which in reaction lose an equivalent amount of momentum? The power-plant only needs to power the M-Drive to do whatever it does. It doesn't have to entirely account for all of the momentum change - part of that can come from the bodies in the star system.