Add support for temporal data

[AdeelH]

Overview

This PR allows Raster Vision to read temporal data i.e. sequences of images.

Major changes:

• New TemporalMultiRasterSource that stacks (rather than concatenates (like MultiRasterSource)) images from sub sources to get spatiotemporal chips of shape (T, H, W, C).
• XarraySource can now accept and read from a DataArray with a time dimension if temporal=True.
• MultiRasterSource can now work with sub raster sources that return >3D chips.
• RasterTransformers can now work with >3D chips.
• RasterStats can now work with >3D chips.
• AlbumentationDataset (base class of GeoDatasets) can now read from a dataset that returns 4D chips.
• Visualizers can now plot batches of 4D chips.
• New demo notebook: Working with time-series of images

Checklist

• Added needs-backport label if PR is bug fix that applies to previous minor release
• Ran scripts/format_code and committed any changes
• Documentation updated if needed
• PR has a name that won't get you publicly shamed for vagueness

Notes

Even though RV GeoDatasets can now return temporal data, training a model on this kind with the default Learners will not work because the default models are image-level models. Therefore, the user will need to provide custom models that work with sequential data.

Testing Instructions

• See updated/new unit tests and new demo notebook

Closes #1802

[codecov]

Codecov Report

Merging #1803 (0d5761e) into master (ba2dc8a) will increase coverage by 0.29%.
The diff coverage is 91.83%.

@@            Coverage Diff             @@
##           master    #1803      +/-   ##
==========================================
+ Coverage   79.53%   79.83%   +0.29%     
==========================================
  Files         187      188       +1     
  Lines        9038     9182     +144     
==========================================
+ Hits         7188     7330     +142     
- Misses       1850     1852       +2     

Impacted Files	Coverage Δ
...astervision/core/data/label_source/label_source.py	75.00% <50.00%> (ø)
...data/label_source/object_detection_label_source.py	83.63% <50.00%> (ø)
.../core/data/raster_transformer/stats_transformer.py	77.14% <50.00%> (ø)
...rvision/core/data/raster_source/rasterio_source.py	79.85% <81.81%> (-0.43%)	⬇️
...ervision_core/rastervision/core/data/utils/misc.py	81.55% <88.23%> (-1.28%)	⬇️
...data/raster_source/temporal_multi_raster_source.py	89.36% <89.36%> (ø)
.../rastervision/pytorch_learner/dataset/transform.py	95.12% <90.47%> (+0.67%)	⬆️
...tervision/core/data/raster_source/raster_source.py	91.04% <92.30%> (+7.17%)	⬆️
...tervision/core/data/raster_source/xarray_source.py	92.23% <93.54%> (-0.54%)	⬇️
...astervision_core/rastervision/core/raster_stats.py	98.75% <94.11%> (-1.25%)	⬇️
... and 13 more

... and 1 file with indirect coverage changes