Add test coverage for UME

[etherealsunshine]

Description

Improves test coverage for the UME model class. Addresses #80

What's added

• test_malformed_smiles_behavior - Documents how UME handles invalid SMILES strings
• test_malformed_protein_behavior - Tests UME's robustness with invalid protein sequences

Features

• Edge case documentation: Tests 11 types of malformed SMILES (unclosed parentheses, invalid characters, empty strings, etc.)
• Protein sequence validation: Tests 10 types of invalid protein inputs (invalid amino acids, special characters, mixed case, etc.)

Usage

# Run the new robustness tests
pytest tests/lobster/model/test__ume.py::TestUME::test_malformed_smiles_behavior -v
pytest tests/lobster/model/test__ume.py::TestUME::test_malformed_protein_behavior -v

Type of Change

• Bug fix
• New feature
• Documentation update
• Performance improvement
• Code refactoring

Testing

• Tests pass locally
• Added new tests for new functionality
• Updated existing tests if needed

Checklist

• Code follows style guidelines
• Self-review completed
• Documentation updated if needed
• No breaking changes (or clearly documented)

[etherealsunshine]

Hi @ncfrey, I've added some tests addressing #80, considering some of the issues were addressed in existing PR's and added RDKit to the list of dependencies since it had probably been missed out. Let me know if this looks good!

[karinazad]

I'm actually working on making Nathan's detect_modality the default option in the UME tokenizer if no modality is specified, so these malformed inputs will be a good stress test for that as well

[etherealsunshine]

let me know if these changes look good!