Auto Selfcal Version 1.0

Following many substantial bug fixes and improvements we're pleased to release a first version of auto_selfcal. This will operate on single-pointing ALMA and VLA data.

Major improvements:
-Support for ALMA spectral scan data
-Fixes to support spectral window IDs that change between MSes desired for self-calibration
-Fixes to the sensitivity calculation
-Improvements to auto-masking thresholding to avoid over-masking
-Use of wproject gridder for VLA L and S-bands
-Fixes for cont.dat file handing
-uvcontsub script generation supports old and new uvcontsub inputs
-improved metadata handling
-full tclean run for post-solint images
-amplitude selfcal enabled (can be disabled with a variable setting)
-Dynamic switching to nterms=2 imaging for data with fractional bandwidth < 0.1 based on S/N.

The self-calibration routines within the ALMA and VLA imaging pipelines in CASA 6.5.4-9-pipeline-2023.1.0.124 closely match the features in this release.

Final phase-only release

This release incorporates a number of new features to enhance selfcal reliability and optimization.

• Several fall backs are implemented for the inf_EB pre-applied gaintable, combine_spw, spwmapping, and no fallbacks
• Option to not do inf_EB pre-apply is removed
• gaintable filenames are now also sanitized to ensure renaming in the inf_EB stage doesn't fail
• if any target has a S/N > 500 at the start of selfcal or at any time during selfcal, nterms is increased to 2
• If using a cont.dat file, commands to perform uvcontsub with the same line emitting regions excluded are written to a script
• Field table names are now cross referenced with fields from listobs to ensure long field names do not get garbled as they sometimes can be with listobs
• MPI environment of CASA will be detected, if started with mpicasa parallel defaults to True, if not parallel defaults to False
• Tested with CASA version up to 6.5.0

Initial_auto_selfcal release

This release of auto_selfcal is relatively stable for both ALMA and the VLA. This version does not use pre-applied gaintables and always uses combine='spw'. This code should operate safely on most data.