The solid parts e.g. the rook have the major bubble flaws on the rear facing surface. The last print I tried before bringing the printer in doors was a solid version of the cockpit canopy that failed at about 50%. Again the the major bubble defects were on the rearward face.
Solid Canopy, 0.1mm LH, clear v1 resin
Rear / Hinge Side
Rear Face
Front Face
@EvanFoss, I have now printed the test cross at the other two layer heights 0.05mm & 0.025mm. It’s my opinion that the rear facing surfaces get progressively worse the as the layer height is reduced. To me this make perfect sense as at 0.05mm the laser flare covers the same patch of resin twice a many time as it does at 0.1mm. At 0.025mm, this increase too 4 time… I hope that makes sense 
I agree with reported comments from Formlabs support that orientating the print at 45 degrees will improve the surface quality as the laser flare path won’t cover the same patch of resin as many time. But from my experience with other parts printed since, you are still get part cured resin in your tank that can only be filtered out. The test piece however, was design and orientated to purposely to test @KevinHolmes theory, which I think it does very well.
Grey resin is probably a better colour for these tests as its quite hard to photograph clear parts.
Test Prints
Printed at 0.1mm Layer height, clear v1 resin
Printed at 0.05mm Layer height, clear v1 resin
Printed at 0.025mm Layer height, clear v1 resin