Unify test_run_transformers* and test_eval_model* scripts

[OliviaLynn]

1. Unifies:

• all combinations of test_run_transfomers.py (_DC2.py, _DC2_redshift.py, etc) into any_test_run_transformers.py
• all combinations of test_eval_model.py into any_test_eval_model.py

2. Combinations can be accessed via new --use-dc2 and --use-redshift flags.

3. $bash run_all.sh runs all combinations (including swin/mvitv2) and outputs to run_all.log

Other:

• Removing old, non-solo scripts as well.
• Skipping train_* scripts, as these do not run as-is at the moment.
• Planning to follow this with a PR to pull some of the config values out into the config files.
• For Code re-org: unify the train scripts #17

[codecov]

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (7839339) 10.76% compared to head (95146b3) 10.85%.

Files	Patch %	Lines
src/deepdisc/astrodet/astrodet.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #98      +/-   ##
==========================================
+ Coverage   10.76%   10.85%   +0.08%     
==========================================
  Files          25       25              
  Lines        3872     3877       +5     
==========================================
+ Hits          417      421       +4     
- Misses       3455     3456       +1     

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

[drewoldag]

Not a blocker - I'm not immediately seeing why this line was added.

[OliviaLynn]

Wasn't sure if I should explicitly mention it (nice eye there)!

It should have been added in the process of combining various versions of train_mapper_cls and moving it into astrodet. The previous versions had all lived in places that had this import, and I had overlooked that in the astrodet PR.

It comes up way below on L1675 and onwards as, say, augs = detectron_addons.KRandomAugmentationList(

[drewoldag]

This all looks pretty reasonable to me. Thanks for putting it together!

[grantmerz]

I think rather than using conditional statements in the script it would be better to place these conditional variables in a solo config. Mostly to better keep track of changes and minimize the clutter of the actual train script (I now see you've planned this in a follow-up PR)