Wrap alpha estimates and out-of-bounds quality control to the range [-pi pi)

[dahonegger]

This fix wraps the difference between estimated and seed alpha to the range [-pi pi) before applying out-of-bounds quality control. The fix is needed because alpha estimates are initialized (in kAlphaPhiInit.m) to the range [-2pi 0] and estimated (via the least squares fit in csmInvertKAlpha.m) in an unwrapped range, but then good alpha estimates are assumed to all lie within pi/2 radians of the seed. This fix also wraps the alpha estimates themselves so that the estimates lie in a standard, wrapped range.

[KateBrodie]

@dahonegger @mpalmsten @RobHolman

Hi Dave, thanks so much for submitting this! Would you mind resubmitting this pull request to merge into the development branch instead of master? As a policy we like to branch off of development, test and merge there, and only merge into master once we know it's all good (at least this is what we try to do :) ) . If you resubmit the pull request I will test it this weekend and provide my feedback/review to the group and then merge if it all looks good and the group approves!

[dahonegger]

Thanks for maintaining good practices! I have resubmitted to the development branch and have closed this one to master. I'm also happy to discuss the edits in more detail, of course!