Bring source-line coverage for cudf-polars interpreter to 100%

[wence-]

Description

Builds out coverage so all lines are at least touched (if not with all dtypes). Fixing a few small bugs.

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[wence-]

I can spin this out into tiny chunks if required, please say before reviewing in detail if that is preferred.

[vyasr]

I can spin this out into tiny chunks if required, please say before reviewing in detail if that is preferred.

Yes please :D

[wence-]

I can spin this out into tiny chunks if required, please say before reviewing in detail if that is preferred.

Yes please :D

Done as:

• Add test coverage for slicing with "out of bounds" negative indices #15990
• Remove mapfunction nodes that don't exist/aren't supported #15991
• Raise early on unhandled PythonScan node #15992
• Fix implemention of any, all, and isbetween #15993
• Test behaviour of containers #15994
• Add full coverage of utility functions #15995
• Add tests covering magic methods of Expr objects #15996
• Add full coverage for whole-frame Agg expressions #15997
• Coverage of binops where one or both operands are a scalar #15998
• Extend coverage of groupby and rolling window nodes #15999
• Fix typo bug in gather implementation #16000
• Remove deprecated ExtContext node #16001
• Add coverage for both expression and dataframe filter #16002
• Add basic tests of dataframe scan #16003
• Add coverage selecting len from a dataframe (number of rows) #16005
• Add test that diagonal concat with mismatching schemas raises #16006
• Add tests of implemented StringFunctions #16007
• Add tests of expression-based sort and sort-by #16008
• Delete unused code from stringfunction evaluator #16032