opt: lock operator is considered a mutation operator

[DrewKimball]

The optimizer currently returns true for IsMutationOp when called on a Lock operator. This results in some differences in behavior for serializable vs read committed, since read committed can plan a lock operator when serializable performs locking in the scan.
Serializable:

root@localhost:26257/defaultdb> SELECT * FROM (WITH foo AS (SELECT * FROM xy FOR UPDATE LIMIT 1) SELECT 1);
  ?column?
------------
         1
(1 row)

Read committed:

root@localhost:26257/defaultdb> BEGIN ISOLATION LEVEL READ COMMITTED;
BEGIN

Time: 0ms total (execution 1ms / network 0ms)

root@localhost:26257/defaultdb  OPEN> SELECT * FROM (WITH foo AS (SELECT * FROM xy FOR UPDATE LIMIT 1) SELECT 1);
ERROR: WITH clause containing a data-modifying statement must be at the top level
SQLSTATE: 0A000

Jira issue: CRDB-48092

[blathers-crl]

Hi @DrewKimball, please add branch-* labels to identify which branch(es) this C-bug affects.

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}