Fix full dataset download after partial

[brownj85]

Context:
Link to shortcut story

If you call load and download only a partial dataset (eg. load("qchem", molname="H2", attributes=["hf_state"])), then try to load the full dataset (call load() again but without the attributes kwarg), it will not download the rest of the dataset. This is because if _download_dataset sees that there is already a file for some dataset, it assumes it has the data it needs. You can add force=True to ensure it's re-downloaded, but you shouldn't need to.

Description of the Change:
If all attributes are requested in qml.load() (e.g attributes is None), and the dataset already exists locally, _download_partial() will be used to download the missing attributes. If force is specified, the existing attributes in the local dataset will be overwritten.

_download_full() will only be called if the dataset does not exist at all locally.

Benefits:
Mixed calls to qml.load() work as expected.

Possible Drawbacks:

Related GitHub Issues:

[github-actions]

Hello. You may have forgotten to update the changelog!
Please edit doc/releases/changelog-dev.md with:

• A one-to-two sentence description of the change. You may include a small working example for new features.
• A link back to this PR.
• Your name (or GitHub username) in the contributors section.

[brownj85]

Heads up, this will be slow on large datasets until #4674 is merged

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (89c2759) 99.64% compared to head (49e8d41) 99.63%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4681      +/-   ##
==========================================
- Coverage   99.64%   99.63%   -0.01%     
==========================================
  Files         377      377              
  Lines       34002    33754     -248     
==========================================
- Hits        33881    33632     -249     
- Misses        121      122       +1     

Files	Coverage Δ
pennylane/data/data_manager/__init__.py	98.65% <100.00%> (+0.12%)	⬆️

... and 42 files with indirect coverage changes

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

[timmysilv]

looks good! just a few things.

[timmysilv]

looks great! you should add yourself to the contributor list too 😄

EDIT: I see you added yourself in the other PR, so nvm :p

[DSGuala]

Looks good to me 👍