🎨 Self-contained definition of CellxGene schema

[falexwolf]

Is part of a sequence of PRs that refactors the curators:

• 🚚 Bulk-rename old-style curators to CatCurator #2413
• 🚚 Move PertCurator from wetlab here and add CellxGene Curator test #2408
• 🚚 Move CellxGene Curator from cellxgene-lamin here #2403
• ✨ Schema-based curators: DataFrameCurator #2388

---

This PR simplifies the definition of constraints to validate against the CxG schema:

Before	After
It was implicitly defined by the state of another database instance passed via the using_key parameter to AnnDataCurator.	It is now self-contained in the CellxGeneAnnDataCurator class.

It also brings down the run time of the CxG test further down from 50s to 20s:

============================= slowest 50 durations =============================
37.28s call     tests/curators/test_pert_curator.py::test_pert_curator
34.08s call     tests/curators/test_cat_curators.py::test_spatialdata_curator
19.95s call     tests/curators/test_cxg_curator.py::test_cxg_curator
6.86s call     tests/curators/test_cat_curators.py::test_df_curator
6.01s call     tests/curators/test_cat_curators.py::test_soma_curator
5.71s call     tests/curators/test_cat_curators.py::test_soma_curator_genes_columns

Collateral:

• I'm removing the _check_valid_keys() utils function because throwing an error upon Curator init upon missing columns/features is problematic: we do want to be able to validate datasets with missing columns and then receive this as a SchemaError during .validate(). This is a common type of error.
• Fixed the signatures for Feature, ULabel & Schema to now include type and is_type.

[falexwolf]

I don't know who wrote this, so FYI @Zethson @sunnyosun -- I don't think this makes sense for the reason defined in the comment

We definitely want to be able to pass datasets with missing columns to the Curator constructor without raising an error