[SPARK-53124][SQL] Prune unnecessary fields from JsonTuple

[wangyum]

What changes were proposed in this pull request?

This PR enhances the GenerateOptimization rule in Spark SQL Catalyst to improve the pruning of unnecessary fields from JsonTuple generators.

Explicitly handled these cases:

• No generator outputs used: remove the Generate node.
• Some generator outputs used: prune the JsonTuple to keep only necessary fields.

Example:

SELECT f2
FROM (SELECT '{"f1": "Spark", "f2": 2025, "f3": 8}' AS json) test
LATERAL VIEW json_tuple(json, 'f1', 'f2', 'f2') jt AS f1, f2, f3

Before this PR:

== Optimized Logical Plan ==
Project [f2#2]
+- Generate json_tuple({"f1": "Spark", "f2": 2025, "f3": 8}, f1, f2, f2), false, jt, [f1#1, f2#2, f3#3]
   +- OneRowRelation

After this PR:

== Optimized Logical Plan ==
Generate json_tuple({"f1": "Spark", "f2": 2025, "f3": 8}, f2), false, jt, [f2#2]
+- OneRowRelation

Why are the changes needed?

Prune unnecessary JSON fields, reducing data processing overhead and improving query performance.

Does this PR introduce any user-facing change?

No.

How was this patch tested?

Added unit tests covering all scenarios: no outputs used, some outputs used, and all outputs used.

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

No.

[wangyum]

cc @cloud-fan

[cloud-fan]

we didn't do this for the ExplodeBase case, why is it needed?

[wangyum]

After reading SPARK-21657, we can remove unrequiredChildIndex = Nil here.