bug: Sanitize View Names

[cfahlgren1]

We need to normalize the view names to ensure they can be created. View names such as CC-MAIN-2013-20 for fineweb-edu will fail.

We can also use SELECT * FROM duckdb_keywords(); to get reserved words in DuckDB as well

[cfahlgren1]

Sometimes the datasets server can return a 500 but recover on next request. Maybe we can add a retry as well

[cfahlgren1]

Additionally, some datasets don't have a parquet export / conversion, example. We should show a small label to the user so they know why there aren't views for that specific dataset

[cfahlgren1]

For datasets like this one that don't have parquet conversions we can show a lable.