Allow dropping qualified columns

[ntjohnson1]

Which issue does this PR close?

• Closes Drop columns only allows unqualified names and silently fails otherwise #19548

Rationale for this change

Explanation in the issue. Motivation coming more concretely from datafusion-python apache/datafusion-python#1305 (comment)

What changes are included in this PR?

• Adds the test from the issue to highlight expected behavior
• Expands drop_columns to coerce things into a fully qualified column to support the range of column varieties
• This further adds a helper to extract the table name associated with the dataframe to simplify use of qualified drop columns support
  • This is potentially the most controversial part. I could see a nicer api being df.col(<name>) to match the expr version but then we probably do repeated checks for the underlying table name unless there is some caching somewhere. Maybe that performance impact isn't significant.

Are these changes tested?

Yes some additional tests are provided.

Are there any user-facing changes?

I had to update the drop_columns(&[]) test since the type can no longer be inferred. I'm not sure if that is representative of any actual use cases though since I expect the more common is a vector that might be empty in which case the type would be specified.

It now requires specifying columns with dots in them similar to other places "\"f.col1\"" to disambiguate from "f.col1".