New function that pretty-prints the metadata dictionary to
address
#74

Adding print_metadata function for pretty printing of metadata

New functions:
- get_weights(key): Retrieve complete weight metadata for a dataset
- get_weight_names(key): Get just the list of weight names
- get_all_weights_for_release(release_name): Get weights for all datasets in a release

- Added a new `GET /weights/dsids/{dsids_str}` endpoint to the PMG Weights API to allow retrieving weights in bulk for multiple datasets at once.
- Refactored `get_all_weights_for_release` in `metadata.py` to batch-process datasets (in chunks of 500) using the new bulk endpoint, significantly reducing API round-trips.
- Updated dataset extraction to correctly poll the `/datasets` API directly instead of strictly relying on local cache, ensuring offline/non-current releases load correctly.
- Added a `tqdm` progress bar to provide visual feedback during bulk weight fetching.
- Expanded the `pytest` mock API handling and added new unit test cases (handling mismatched tag fallbacks, 404 errors, missing e_tags, and malformed datasets) to achieve 100% test coverage for `metadata.py`.

See:
https://gitlab.cern.ch/atlas-outreach-data-tools/pmg-weights-db-api/-/commit/3d66dfb3631c01d1468276983caa53c49c009900

Better management of modules, less confusion and shorter files.

In addition, Zach's cosmetic comments

feat: weights implementation for atom