[SPARK-56030] [SQL] Normalize inner Project when below Project -> Filter

[mihailoale-db]

What changes were proposed in this pull request?

In this PR I propose to normalize inner Project when below Project -> Filter.

Why are the changes needed?

Right now

SELECT a.key, b.key AS key_b
FROM t1 b
FULL JOIN t2 a USING (key)
WHERE b.key IS NULL OR a.key IS NULL

throws LOGICAL_PLAN_COMPARISON_MISMATCH for analyzer dual-runs and normalizing would fix it.

Does this PR introduce any user-facing change?

No.

How was this patch tested?

Added + existing tests.

Was this patch authored or co-authored using generative AI tooling?

Yes.

[mihailoale-db]

@dtenedor PTAL when you find time. Thanks!

[mihailoale-db]

@cloud-fan PTAL when you find time. Thanks!