feat: add datacompy compare CLI with multi-backend support

[fdosani]

Implements the datacompy compare CLI (closes #530), enabling dataset comparisons from the shell and CI/CD pipelines without writing Python.

• Adds datacompy/cli/ subpackage: parser.py, compare.py, backends.py, loaders.py, sessions.py, output.py, errors.py
• Supports all four backends: polars (default), pandas, spark, snowflake
• Loads CSV, Parquet, and JSON files; Snowflake can compare native table refs (db.schema.table) or stage local files to a temp table
• Exit codes: 0 match, 1 mismatch/threshold violated, 2 error
• --max-unequal-rows N for threshold-based CI assertions
• --json for machine-readable output; --quiet for exit-code-only pipelines
• --debug re-raises unexpected exceptions for bug reporting
• Argument validation split between argparse type= callables (value rules) and _validate_arg_combinations (cross-argument rules)
• Session lifecycle managed via contextlib.ExitStack — Spark and Snowflake sessions close on both success and failure
• All optional backend imports are deferred so datacompy --version does not pay the pyspark/snowflake import cost