Check qpoint treatment of el > 90

[mhasself]

As reported by @amaurea, qpoint doesn't seem to handle el > 90 totally cleanly -- likely due to refraction correction.

We should assess how this affects so3g usage and guard against it (and notify upstream if warranted).

[arahlin]

I'm happy to fix this upstream and make a new qpoint release if that's easier.

[mhasself]

I'm happy to fix this upstream and make a new qpoint release if that's easier.

I think that would be very helpful @arahlin -- let me know if you need more details. I've now confirmed the issue is related to refraction

A simple fix in the so3g level (which I will pursue soon in absence of qpoint patch) is confirmed to be:

• before calling qpoint, transform (az, el) -> (az + 180, 180 - el) for the affected samples.
• after calling qpoint, correct for focal plane flipping by right-applying euler(2, 180*DEG) for the affected samples.

[arahlin]

Should be straightforward to implement.

[arahlin]

I think arahlin/qpoint#18 is the appropriate solution here. Would you be willing to test the fix?

[arahlin]

Ok, I've released qpoint version 1.13.0 on PyPI with these changes.