[CORE] Bad mutable query plan nodes

### Description

Spark / Gluten query planner requires heavily on plan immutability.

Currently there are some (unlimited to) code snippets potentially making plans mutable:

1. WholeStageTransformer.setOutputSchemaForPlan (CH) https://github.com/apache/incubator-gluten/blob/15f4cde02bf965e70bae5079c40aae78ad80e45a/gluten-substrait/src/main/scala/org/apache/gluten/execution/WholeStageTransformer.scala#L234-L241
2. BatchScanExecTransformer.setPushDownFilters (VL + CH) https://github.com/apache/incubator-gluten/blob/15f4cde02bf965e70bae5079c40aae78ad80e45a/gluten-substrait/src/main/scala/org/apache/gluten/execution/BatchScanExecTransformer.scala#L102-L104

Should find a way to optimize against the code to make relevant plan immutable.

Start by reading WholeStageTransformer.setOutputSchemaForPlan and BatchScanExecTransformer.setPushDownFilters at the linked revisions. Trace how these methods affect Spark/Gluten query plan nodes for the CH, VL, and CH paths. Done means the relevant plan nodes remain immutable while preserving the required planner behavior.

Written by the indexing model from the issue text.