fix(cli): infer delimiters for TSV inputs

[AtomicGlance]

Addresses #545.

What changed

• Infer comma or tab delimiters separately for each input file.
• Recognize .tsv and .tab as CSV inputs across pandas, Polars, and Spark.
• Keep --csv-delimiter as an override for both inputs.
• Replace the old TSV limitation test with same-format and mixed CSV/TSV coverage, including a Spark regression test.
• Update the CLI documentation to describe the inferred extensions.

Why

The CLI already inferred each file's format independently, but the delimiter was one shared value that defaulted to a comma. A TSV file could therefore reach the correct reader and still be parsed into one malformed column. Keeping the parser default unset lets the loaders distinguish an explicit override from normal extension-based inference.

Validation

• pytest tests/cli/test_compare.py tests/cli/test_parser.py tests/cli/test_spark.py -q — 124 passed, 5 skipped
• ruff check datacompy/cli tests/cli
• ruff format --check datacompy/cli tests/cli
• mypy --follow-imports=skip --ignore-missing-imports datacompy/cli/backends.py
• sphinx-build -b html docs/source docs/build/html (builds with the existing api/modules toctree warning)

[CLAassistant]

All committers have signed the CLA.

[fdosani]

Thanks for the PR. I'd like to hold off merging this in right now as I want to focus on getting the original PR reviewed and merged. I'll leave it be for now, but maybe get you to do a rebase/alteration once we are ready to move forward with this.

[fdosani]

Taking another pass at the review. Sorry about the confusion