ENH #61033: Add coalesce_keys option to DataFrame.join

This adds a coalesce_keys keyword to DataFrame.join to allow
preservation of both join key columns (id and id_right),
instead of automatically coalescing them into a single column.

This is especially useful in full outer joins, where retaining
information about unmatched keys from both sides is important.

Example:
    df1.join(df2, on=id, coalesce_keys=False)

This will result in both id and id_right columns being preserved,
rather than merged into a single id.

Includes:
- Modifications to join internals (core/reshape/merge.py)
- A dedicated test file (test_merge_coalesce.py) covering:
    - Preservation of join keys when coalesce_keys=False
    - Comparison with default behavior (coalesce_keys=True)
    - Full outer joins with asymmetric key presence

Co-authored-by: Maria Pereira <[email protected]>

Author: rit4rosa