I refined the collimation a bit and I think I'm on the right track. There's still some egg shape in the corners, mostly at the top of the frame.

The night sky stabilised nicely after the collimation effort so I took 3 x 5min test subs (1x1 binning) of the vicinity of NGC 3263 to check star shapes. The seeing was average. This image is derived from this data. Each box is a full resolution patch of sky 2.3 x 1.7 arc minutes in size (ie a very small patch of sky). Image scale is 0.59 arc second/pixel. The boxes represent the extremities and middle of the full frame. The 3 subs I took were dark subtracted, sum combined and stretched. Flats were applied but there was no deconvolution or sharpening of any kind.

Such raw results are very good!