neoline361-art/fitcheck ? reverse-engineered prompt

Build me a simple Python tool called FitCheck that helps me quickly check ML data and models without any setup headaches.

I want to point it at a CSV or Parquet file and get back a clear report that shows basic data quality issues like missing values, duplicates, constant columns, class imbalance, and outliers. I also want to compare two datasets and see if anything has drifted between training data and production data. If I already have a model, I want to run an evaluation and get useful metrics and plots in a friendly HTML report.

Please make it work from both Python and a command line, so I can run something like a quick check, a drift comparison, or a demo without writing much code. Keep it local only, no telemetry, and make the reports easy to share by email or Slack. If you need to look up current docs online for anything, go ahead.