question on printing thin walls at high res

having nailed a 0.1mm profile I decided to have a try at a 0.5.

I've got the speed settings sorted, and found that a substantial skirt/brim is needed to stop the 1 walled flow calibration cube from lifting straight away. (at the moment, I've got parts sticking to the bed, now I want to get the walls as thin as possible.)

but now I'm having a new problem.
ignore the fact that the bottom is warped - that only happened because I pulled it off the bed whilst it was still at 100degrees.

I've got a case and door. the enclosure temperature is measured at 48 degrees C.

bed temperature is set to 100, (and is reaching it) I have a 1.7mm sheet of glass, (so it's not big and leaching loads of heat from the bed) I've got kapton on top of the glass to aid sticking.

the results of 0.05mm printing are good (in the parts that work) you can't actually see any lines at all, what look like lies are actually more like little squishes of filament where the extrusion multiplier is set too high, (hence why I'm printing a measurable object to correct that...

Yes, the gauge is still in that photo, you just can't make it out in profile....

this video explains that more.

you can see, that you can't see lines, any thinker than the gauge that's shown for comparison.
a gauge so thin that you can't see it in profile with the camera (iphone 4s camera) the gauge is 0.05mm

Anyway, having gotten sticking to the bed sorted,  extruder speed etc (very slow 5mm/s -any faster and sticking is a real problem) and wall thickness is now controlled.
as you can see in the photos the stress that normally lifts parts from the bed has instead torn the very thin walls at the corners.

I know it's the hot plastic/cold part shrinking that's causing this. but what can I do to stop it?