When we model vertical supports Y or +Y we do not typically include moment restraints RX or RZ.
My understanding is that moment restraints would not reflect reality as minor rotations will always occur due to support flexibility and tolerances. i.e. zero angle restraint isn't achievable.

The mechanical engineer designing the supports is insistent there will be a moment at these supports (flat plate on flat plate) and i concur. Clearly without RX or RZ restraints there will be no moments from the model, however if I restrain these rotations to provide him with the moments at this restraint I am concerned that these moments are artificially imposed on the system and will not really happen.

What is a reasonable approach here to feed back moments for the support design? Would it be practicable to decrease the default support stiffness for moment restraints? Again i'm concerned that this would be a bit of a finger in the air approach. (Note that max rotations in the horizontal plane are <0.05 degrees).

Alternatively I suppose I could model the width of the plate (as pipe to model the support stiffness) with vertical supports either side to give a torque. I feel like this is over-complicating things too.

I think a +Y support at each end of the plate might give you what you want. If there is a sufficient moment coming down the line, the pipe can lift off one end or the other.
But, eventually, I believe you will find this tedious and of no benefit to good analysis.