Summary of changes in the 0.18.x release

Updates to Level-2a processing for all exposure modes

Fixed bug in the minimum number of groups per integration for the jump detection step by changing it from 3 to 5
Reworked the jump step to reduce memory usage
Updated the ramp_fit step to add 'DO_NOT_USE' DQ flag to pixels that have all groups flagged as saturated
Turned the MIRI "reset" step back on in the level-2a pipeline
Updated the saturation step to use the A-to-D limit of 65535 as the saturation threshold for pixels flagged with "NO_SAT_CHECK" in the saturation reference file
Updated the saturation step to flag groups having values below the A-to-D floor (zero)

Enhance the flat_field step to flag MIRI imager NON_SCIENCE pixels as DO_NOT_USE
Enhancements to assign_wcs to add FITS SIP approximation keywords to imaging mode WCS

Enable the resample_spec step for NIRSpec line lamp exposures
Enhancements and bug fixes to the extract_1d step for extracting and applying aperture corrections to IFU mode data
Bug fixes to the wavecorr step to produce correct wavelength shifts
Enable processing of NRS_LAMP exposures in the assign_wcs and resample_spec steps
Enhancements and bug fixes to cube_build for handling single cubes
Bug fix for extract_1d when determining source RA/Dec in input resampled Slit data
Enhanced extract_1d to use an extract1d reference file to guide processing for NIRCam TSGRISM exposures, allowing for column-by-column fitting and subtraction of background signal
Bug fix for extract_1d when there's only one spectrum instance for a WFSS source
Updated flat_field step branch logic to handle NRS_LAMP as a spectroscopic exposure
Updated NIRSpec fixed-slit processing in the flat_field and photom steps to compute and save correction values for both point and uniform sources in the primary slit when it contains a point source, in order to support master background corrections

Bug fix in outlier_detection to account for background signal when blotting the median-combined image back to individual exposure's space
Bug fix in the skymatch step to prevent failures when attempting to compute statistics on a single pixel
Implemented memory checks in the outlier_detection and resample steps to prevent the creation of huge arrays and subsequent memory allocation errors
Added "pixel_scale_ratio" parameter to the resample step to allow for finer output grid
Added support for the new fitting geometry mode "rshift" in the tweakreg step

Update the NIRSpec MOS master background logic to only proceed with processing after verifying that there are both background and source slits available in the input dataset
Enhancement to the combine_1d step to produce separate extensions for each spectral order

Updated the tso_photometry step to place the photometric aperture at the location given by the XREF_SCI, YREF_SCI keyword values, instead of using CRPIX1, CRPIX2
Fixed the flux units in the source_catalog output
Updated the white_light step to allow for the specification of wavelength limits to restrict the portion of the spectrum used to compute the integrated flux

Complete overhaul of the ami_analyze step to use the latest ImPlaneIA algorithms from the AMI WG
Enhancements to the ami_analyze step to allow two user-settable parameter

Update documentation introduction to include installation and CRDS setup
instructions. [#5659]

Fixed code error in combine1d, creating extensions per spectral order
with the same input data [#5644]

Fix a bug in estimating the max number of segments that will be needed
to fit any pixel [#5653]

Update the check in set_telescope_pointing that determines whether an
exposure is TSO mode to always consider hardwired TSO EXP_TYPEs as TSO,
regardless of TSOVISIT and NINTS settings. [#5657]