Reconcile chip-sampling functionality in RVPipelines with GeoDatasets (#1496)

[AdeelH]

Overview

This PR unites the code paths for direct chip sampling and chip sampling during the chip stage. The chip stage now uses GeoDatasets to sample chips, while still retaining the ability to filter out chips based on their content (e.g. based on NODATA %).

The window sampling configs have also been united such that both ChipOptions and GeoDataConfig now accept the same config: WindowSamplingConfig.

See commits for more details.

Checklist

• Added unit tests, if applicable
• Updated documentation, if applicable
• Added needs-backport label if the change should be back-ported to the previous release
• PR has a name that won't get you publicly shamed for vagueness

Notes

Despite the major changes, backward compatibility with older model bundles has been retained. This required enhancing Registry and upgrade_config() with the ability to handle config renamings.

Testing Instructions

See updated unit and integration tests.

Closes #1496

[codecov]

Codecov Report

Attention: 39 lines in your changes are missing coverage. Please review.

Comparison is base (2ed63e3) 86.35% compared to head (0095fa6) 89.46%.

Files	Patch %	Lines
...n/core/rv_pipeline/semantic_segmentation_config.py	71.42%	10 Missing ⚠️
...astervision/core/rv_pipeline/rv_pipeline_config.py	42.85%	8 Missing ⚠️
...ner/rastervision/pytorch_learner/learner_config.py	68.75%	5 Missing ⚠️
...rvision/pytorch_backend/pytorch_learner_backend.py	92.30%	3 Missing ⚠️
...vision/pytorch_backend/pytorch_object_detection.py	90.00%	3 Missing ⚠️
...tervision_pipeline/rastervision/pipeline/config.py	85.71%	2 Missing ⚠️
...ision/pytorch_learner/regression_learner_config.py	77.77%	2 Missing ⚠️
...ion/pytorch_backend/pytorch_chip_classification.py	95.00%	1 Missing ⚠️
...n/pytorch_backend/pytorch_semantic_segmentation.py	95.23%	1 Missing ⚠️
...n/pytorch_learner/classification_learner_config.py	88.88%	1 Missing ⚠️
... and 3 more

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2040      +/-   ##
==========================================
+ Coverage   86.35%   89.46%   +3.11%     
==========================================
  Files         196      198       +2     
  Lines        9760     9729      -31     
==========================================
+ Hits         8428     8704     +276     
+ Misses       1332     1025     -307     

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