Feature/add unit tests

[sray014]

This PR makes the following changes to ras-stac:

• Adds unit tests for stac item creation functions
• Moves reading of s3 hdf files outside of item creation functions
• Combines ras_hdf and ras_stac into one ras_utils file
• Organizes item creation functions into 2 classes, one for geom files and one for plan
• Converts all attribute times into iso format instead of datetime.datetime, removing the need for datetime json encoder

[mxkpp]

In addition to the per-file comments and per-line comments included elsewhere in this review...

Additional comments required to be resolved before merge:

• Docker setup: Dockerfile fails to build since ras_stac/plugins does not exist.
• Docker setup: Add a .dockerignore file and use it to explicitly avoid including .env into the image. Even though this is currently not an issue (Dockerfile does not have a COPY . . directive), might as well do it.

Additional comments that could be made into issues if not resolved prior to merge:

• DRYify Logging: Start a new file ras_stac/utils/logger.py with a function set_up_logger that each module and script can import, then consolidate the various logging calls.
• Idiomacy: When a variable holds an S3 path / URL having the form "s3://bucket/some/great/blob" Use var name s3_path rather than s3_key or s3_full_key
• Explicit failures when environment is not set up properly: Use os.environ["VARNAME"] instead of os.environ.get("VARNAME") since the former will fail with a clear KeyError if the env var has not been set.
• Stability: Consider upgading version of GDAL base image in Dockerfile (currently 3.8.0) to incorporate the latest bug fix releases within 3.8.x.
• Robustness: Near the top of populate-sample-data.sh, modify existing line set -e to be the standard full set -euo pipefail. Also move this line to the very top, just underneath the shebang, before the source .env.
• Convenience / Ease of Use: In populate-sample-data.sh, invoke the Python runtime like python -um ... instead of python -m .... The u causes Python to run unbuffered, so you can see lines in realtime as they are printed/logged rather than waiting for a large chunk of lines to flush.

[mxkpp]

Thanks for the updates. I just pushed a small change to an error message, good to go.

Edit: I did push one more commit to add core tests into the Docker image, and to add a shell script for building the Docker image and running those core tests in the container.