Update AMI NRM geometry

[stscijgbot-jp]

Issue JP-3739 was created on JIRA by Rachel Cooper:

For build 11.2, update the use of the NRM reference file in AmiAnalyze.

• Replace hardcoded mask hole centers, other pupil geometry constants with those in NRMModel. Replace mask_definitions.py with mask_definition_ami.py, removing references to NRMs on other telescopes and simplifying the code but retaining the same structure as the existing class.
• A special input argument for affine2d, 'commissioning', will be the new default, using the commissioning affine parameters for the analytical form of the mask (rather than distortion-free or rotation-only).
• Rename the lg_model.NrmModel to lg_model.LgModel to avoid confusion with stdatamodels' NRMModel.

[stscijgbot-jp]

Comment by David Law on JIRA:

Rachel Cooper What exactly is involved in this?  It sounds like hole coordinates are hard-coded in jwst, is the intent to replace them with use of reference files (new or just updated?) or change the hard-coded values?  Are the new values known and ready to go?

[stscijgbot-jp]

Comment by Rachel Cooper on JIRA:

Hi David Law, currently the hole coordinates are indeed hard-coded in AMI3 ([https://github.com/spacetelescope/jwst/blob/master/jwst/ami/mask_definitions.py)] but they are also duplicated in the header of the recently created NRM reference file, so I think it would probably be better practice to read them from the reference file instead. First I need to calculate the updated coordinates using the distortion matrix that was derived during commissioning, then update the existing reference file with a mask model made using those coordinates (and store them in the header), and then update the jwst code to use those values rather than the mask_definitions.py ones. It might take a little thought as to the best way to modify the code to do this just due to how it's currently set up, but I don't think it will be too involved.

[stscijgbot-jp]

Comment by David Law on JIRA:

Assigning to the NIRISS team to sort out final details, though it sounds like that should be straight forward.  Rachel Cooper I concur that reading coordinates from the reference file would be better practice than hard-coding duplicates that can drift out of sync.  Once everything is set on the NIRISS side we can loop in Tyler Pauly to discuss best path implementation.

[stscijgbot-jp]

Comment by Rachel Cooper on JIRA:

Draft PR opened here: #8974 Will open it for review after #8846 (JP-3718) is merged. 
I re-ran our AMI science readiness commissioning analysis and found that just updating the mask did not make a significant change in the contrast limits, however it does reduce the maximum residual in the image plane model fitting.