Delay materializing RangeIndex in .reset_index

[mroeschke]

Description

Before, a RangeIndex would be materialized to a Column even if it wasn't used (drop=True). Now, it's only materialized if the index will be added as a new column in the DataFrame.

Also caught a validation bug where an invalid number of levels would not raise an error

Checklist

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

[galipremsagar]

Nothing jumps out for me while looking at the code but @mroeschke can you look into why there are new failures in the following files:

• tests/generic/test_finalize.py
• tests/frame/methods/test_to_period.py
• tests/frame/methods/test_rank.py

It might be noise but wanted to be sure if these failures are new because of this PR or not.

[mroeschke]

can you look into why there are new failures in the following files:

Did you see these failures locally? Looks like the CI didn't pick up these failures

[vyasr]

@galipremsagar could you comment on where you saw those failures?

[galipremsagar]

@galipremsagar could you comment on where you saw those failures?

They were here: https://github.com/rapidsai/cudf/actions/runs/8823446932/attempts/1#summary-24226360963

But after looking at my PRs, these failures are probably un-related. So ignore my comment.

[vyasr]

Based on Prem's comment above, approving assuming that my understanding in the one question that I raise is correct.

[mroeschke]

/merge