Write a robust Python automation script with error handling

Write a Python script that does the following: [describe the task precisely, e.g. "reads all CSV files in a folder, validates required columns, merges them into one deduplicated CSV, and logs any rows it had to skip"].

Environment:
- Python version: [e.g. 3.11]
- Allowed dependencies: [list any you want to use, or say "standard library only" / "pandas is fine"]
- How it will run: [manually / cron job / CI pipeline / triggered by another script]

Requirements:
1. Use a proper CLI interface (argparse or click) with --help text, not hardcoded paths
2. Validate all inputs before doing real work (file exists, required columns present, correct types) and fail with a clear error message rather than a raw traceback
3. Handle partial failures gracefully — if one file/record fails, log it and continue rather than crashing the whole run, unless I say otherwise
4. Use the logging module (not print) with an adjustable log level
5. Make it idempotent where possible — running it twice on the same input should not corrupt output or duplicate data
6. Add type hints throughout and docstrings on every function
7. Write pytest tests covering: normal input, empty input, malformed input, and at least one failure-recovery case
8. Include a short usage example in a comment at the top of the file

After writing it, tell me what assumptions you made about the input format, and what I should double check before running this against real data.