===================
- Added PSF modelling routines. These are currently very preliminary, but at least exist
- Allow for external, absolute catalog in ``get_wcs_adjust_step``
- Add PSF matching routines (``psf_matching``)
- Add anchoring routines (``anchoring``)
- Include useful outputs from these in the ``release_step``
- If not grouping dithers in ``get_wcs_adjust_step``, respect that in how the transforms are
written out
- ``single_tile_destripe_step`` can now run on rate files (pre-flat fielding)
- Bugfixing in ``single_tile_destripe_step``
- Decoupled horizontal/vertical destriping methods in ``single_tile_destripe_step``, since the
noise properties are distinct in these two axes
- Add ``smooth`` option to ``single_tile_destripe_step``, based on Dan Coe's smooth1overf
algorithm
- Added control over how values are extended beyond array edge for filtering in ``single_tile_destripe_step``
- Lots of bugfixing in ``multi_tile_destripe_step``
- ``multi_tile_destripe_step`` can un-flat before correcting
- Level between amplifiers in ``multi_tile_destripe_step``
- Make vertical stripe subtraction optional in ``multi_tile_destripe_step``
- Added median option to ``multi_tile_destripe_step`` to do a median rather than mean image
- Added iterative option to ``multi_tile_destripe_step`` that will keep things going until
sigma-based convergence
- Added sigma-clip median option for creating stacked images
- ``do_large_scale`` now works completely differently in ``multi_tile_destripe_step``, instead
attempting to clean up the average image
- Added support for different ``do_large_scale`` methods in ``multi_tile_destripe_step``,
which may work better in certain situations
- Added a median filter ``do_large_scale`` method, which may be more robust than the boxcar. THIS
IS NOW THE DEFAULT
- Added a sigma-clipped ``do_large_scale`` method, as should be optimal in observations that aren't
full of emission
- Added a smooth convolution ``do_large_scale`` method, based on Dan Coe's smooth1overf algorithm
- Added control over how values are extended beyond array edge for ``do_large_scale``
- Added option in ``lv3_step`` to degroup dithers for tweakreg
- Added option in ``lv3_step`` to degroup NIRCam modules, since the WCS is currently inconsistent
between the two
- Fixed crash in ``lv3_step`` if one of the group/degroup parameters is not defined
- Be smarter about keeping track of exposure numbers in ``lv3_step``
- ``regress_against_previous`` will now search for files in priority order, for fallback between versions
- Make sure backgrounds are included in label for ``regress_against_previous``
- f-string fixes