Fix Index.difference to match with pandas

[galipremsagar]

Description

This PR fixes Index.difference in following ways:

• Fixes name preservation by correctly evaluating the name of two input objects, closes [BUG] Incorrect name selection in Index.difference when an object with no-name is passed #14019
• Fixes is_mixed_with_object_dtype handling that will resolve incorrect results for CategoricalIndex, closes [BUG] CategoricalIndex leading to incorrect results in Index.difference #14022
• Raises errors for invalid input types, the error messages are an exact match to pandas error messages for parity.
• Introduce a Range._try_reconstruct_range_index that will try to re-construct a RangeIndex out of an Int..Index to save memory- this is on parity with pandas. closes [FEA]: Preserve RangeIndex in more setops cases #14013

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-]

Looks good overall, but I think there's a missing corner-case in the reconstruct logic, and cupy appears to be a bit fast than conversion to a frame for the unique call

[mroeschke]

Does this close #14013?

[galipremsagar]

Does this close #14013?

Yup. Linked to the PR.

[wence-]

Thanks! (Didn't know an index could have nulls!)

[galipremsagar]

/merge