[FEATURE] Data as git submodule

[jesse-wei]

The repo is ~2G when cloned due to data/.

data/ should be in a separate GH repo and added as a submodule to this repo.

Developers can then clone with --depth=1 (and probably --no-single-branch) to avoid initializing data/, if they don't need it. They can clone with --recurse-submodules to initialize data/. More info in git documentation about submodules.

If you choose not to clone data files, then you should use pytest's --ignore option to ignore tests that use files in data/. tests.yml may also have to be configured.

[eric-unc]

skill lack