JP-3567: SIP coefficient accuracy and consistency

[melanieclarke]

Resolves JP-3567

Closes #8341

These changes align the default parameters for WCP SIP approximation in the three places they are used: the assign_wcs step, the tweakreg step, and the underlying update_fits_wcsinfo function defaults. New step parameters are added for tweakreg, matching the ones in assign_wcs, so that the parameters used in both steps can be modified together.

The new defaults for assign_wcs should result in better accuracy in the output FITS WCS.

Checklist for PR authors (skip items if you don't have permissions or they are not applicable)

• added entry in CHANGES.rst within the relevant release section
• updated or added relevant tests
• updated relevant documentation
• added relevant milestone
• added relevant label(s)
• ran regression tests, post a link to the Jenkins job below.
  How to run regression tests on a PR
• All comments are resolved
• Make sure the JIRA ticket is resolved properly

[melanieclarke]

Regression tests started here:
https://plwishmaster.stsci.edu:8081/job/RT/job/JWST-Developers-Pull-Requests/1488/

[codecov]

Codecov Report

Attention: Patch coverage is 75.00000% with 2 lines in your changes missing coverage. Please review.

Project coverage is 58.13%. Comparing base (b7e0b10) to head (66d3f0c).
Report is 311 commits behind head on master.

Files with missing lines	Patch %	Lines
jwst/tweakreg/tweakreg_step.py	60.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8529      +/-   ##
==========================================
+ Coverage   58.02%   58.13%   +0.10%     
==========================================
  Files         388      388              
  Lines       38977    38972       -5     
==========================================
+ Hits        22617    22655      +38     
+ Misses      16360    16317      -43     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[melanieclarke]

Regression tests look mostly as expected. There are some unrelated changes for MIRI LRS; all other changes are to the WCS keywords in the SCI headers for imaging and WFSS modes.

[hbushouse]

@melanieclarke Is this ready for review?

[melanieclarke]

@hbushouse - Yep, I think it's good to go. Nadia's review would be helpful, if she can, since she worked on this ticket.

[hbushouse]

Everything looks good to me. Just curious, though, as to why the regtest results show that for most results the order of the fit (and hence the number of SIP coeff keywords in the headers) has increased, but for one MIRI imaging result it appears that the order of the fit has been reduced compared to before. Any thoughts?

[melanieclarke]

@hbushouse - I saw that. I suspect it's because we increased the number of points, so it needed a smaller order to reach the required accuracy, but I haven't checked that assumption. I'll check it out.

EDIT:
My assumption was wrong. It looks like the WCS fit for the MIRI image is failing to converge to the requested accuracy with the larger number of sampling points. MIRI might want to set their parameters for these steps to a lower value for sip_npoints and/or a higher value for sip_max_pix_error.

[nden]

LGTM

[melanieclarke]

I'm going to push up one more small change to set the default npoints to 12 instead of 32, per discussion in the JP ticket.

[hbushouse]

Another regtest run (hopefully final) started at https://plwishmaster.stsci.edu:8081/job/RT/job/JWST-Developers-Pull-Requests/1508

[hbushouse]

Latest regtest results, while corrupted by lots of unrelated differences in NIRSpec slit data due to updates to wavecorr, show the expected increase in SIP coeff keywords for all relevant modes, such as NRC, NIS, and MIRI imaging, as well as NRC and NIS WFSS. None of them shows a decrease in the fit order anymore. So this looks good.