Can you write a summary conclusion from this work - what is it telling us. Is this saying that phosim doesnt model the camera rotation and so is not useful for testing WFS with a non-zero rotation?

Phosim models camera rotation as well as sky rotation. However, the `rotTelPos`, which is the angle of telescope relative to camera in decimal degrees (also called `camtelangle`) is considered a  "standard" input according to PhoSim documentation https://bitbucket.org/phosim/phosim_release/wiki/Instance%20Catalog . In contrast,   `rotSkyPos` which is the angle of sky relative to camera coordinates (from North over East) in decimal degrees  (also called `camskyangle`), is considered one of the "advanced" inputs (which can be calculated self-consistently with  PhoSim). Calculation of one or the other is invoked here https://github.com/lsst-ts/phosim_syseng4/blob/6105fcf9d65b37d2d157be1f99d70dc263da0e76/source/atmosphere/main.cpp#L315-L332 using the following functions https://github.com/lsst-ts/phosim_syseng4/blob/6105fcf9d65b37d2d157be1f99d70dc263da0e76/source/atmosphere/operator.cpp#L127-L138  . Thus case3 above seems most realistic - in the real world, we can't move the sky around (objects are fixed in the sky at a given moment), we can only slew the telescope (move "left/right" or "up/down" in the alt/az  fork mount like LSST), and then rotate the camera (so that the lower edge of the FOV is not aligned with the horizon, but follows the sky movement, see https://telescopemount.org/alt-az-mounts-for-long-exposure-astrophotography-camera-rotators/ ). 

This shows that the `rotTelPos` command does not fully rotate the camera, but only the spider . The camera is kept at a fixed angle wrt M1M2, and the only thing that rotates due to changing "rotTelPos" is the spider. This means that if we want to model camera rotation with PhoSim, we could  eg.  rotate the calculated grid surface for the mirrors (i.e. rotate the mirrors rather than the camera). 

Looks good - so the only option for phosim is to rotate the perturbations of the mirrors?

Yes, it appears so.