*: fix a bug that the pessimistic lock doesn't work on a partition (#14921)

[sre-bot]

cherry-pick #14921 to release-4.0

---

What problem does this PR solve?

'select ... for update' doesn't work right on a partitioned table.

txn1: begin
txn1:  select * from t where a = xx for update

txn2:  update t set b = yy where a = xx

In pessimistic txn mode, txn2 should be blocked.

What is changed and how it works?

The bug is that we use table ID to construct the lock key.
On a partitioned table, partition ID should be used.

Changes:

• During column pruning, do not prune the partition columns for LogicalLock
• In SelectLockExec, use table ID and partition columns to locate the partition, then use the partition ID to construct the lock key.

Check List

Tests

• Unit test

Code changes

• Has exported function/method change

Side effects

• Possible performance regression
• Increased code complexity

Related changes

• Need to cherry-pick to the release branch

Release note

• Write release note for bug-fix or new feature.

[sre-bot]

/run-all-tests

[zz-jason]

@tiancaiamao Please confirm that whether we should cherry pick PRs one by one from master to release-4.0. I remembered that we are going to fast forward the code from master branch to release-4.0.

[tiancaiamao]

OK... maybe I see someone cherry-pick and following him
The type/4.0 cherry-pick is misleading...