[fix](coordinator) build TPlan within the table lock to prevent NPE in getSchemaByIndexId

[feiniaofeiafei]

What problem does this PR solve?

Fix NPE: Cannot invoke "org.apache.doris.catalog.MaterializedIndexMeta.getSchema()" because the return value of "java.util.Map.get(Object)" is null in OlapTable.getSchemaByIndexId.

Related: DORIS-23676, CIR-20142

Root cause (race condition):

1. Nereids planner acquires table read lock, selects selectedIndexId = baseIndexId, then releases the lock after planning
2. A concurrent SchemaChangeJobV2.onFinished (under write lock) calls deleteIndexInfo(originIdxName) which removes originIdxId from indexIdToMeta
3. Later, OlapScanNode.toThrift() calls getSchemaByIndexId(selectedIndexId) without holding any table lock → returns null → NPE

Fix (same approach as #59298 for master):

1. NereidsPlanner: call cacheThriftPlans() inside the lock callback (while table read lock is still held), serializing the TPlan Thrift under the lock
2. PlanFragment: add cacheThriftPlan() that pre-serializes planRoot.treeToThrift() via a memoized Supplier; toThrift() reuses the cached result
3. OlapTable.getSchemaByIndexId: add null guard that throws a clear RuntimeException with context info instead of an opaque NPE

Release note

None

Check List (For Author)

• Test

  • Regression test
  • Unit Test
  • Manual test (add detailed scripts or steps below)
  • No need to test or manual test. Explain why:
    • This is a port of [opt](coordinator) build TPlan within the table lock #59298 from master to branch-3.1

• Behavior changed:

  • No.

• Does this need documentation?

  • No.

[hello-stephen]

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
3. What features were added. Why was this function added?
4. Which code was refactored and why was this part of the code refactored?
5. Which functions were optimized and what is the difference before and after the optimization?

[feiniaofeiafei]

run buildall

[feiniaofeiafei]

/review

[feiniaofeiafei]

run buildall

[feiniaofeiafei]

/review

[feiniaofeiafei]

run buildall