*: add session variable "tidb_enable_cascades_planner"

[zz-jason]

What problem does this PR solve?

This PR is split from #7869

What is changed and how it works?

• add a session variable tidb_enable_cascades_planner to enable the new cascades planner.
• move Optimize() out of the package plannercore to planner, chose the query planner according to the session variable tidb_enable_cascades_planner

Because the cascades planner is not implemented yet, an error will be returned if we use the cascades planner to process a query:

TiDB(localhost:4000) > select * from t;
+---+------+------+
| a | b    | c    |
+---+------+------+
| 1 | NULL | NULL |
+---+------+------+
1 row in set (0.00 sec)

TiDB(localhost:4000) > set @@tidb_enable_cascades_planner=on;
Query OK, 0 rows affected (0.00 sec)

TiDB(localhost:4000) > select * from t;
ERROR 1105 (HY000): The cascades planner is not implemented yet.

Check List

Tests

• Unit test

Code changes

• Has exported function/method change

[zz-jason]

@eurekaka @winoros @lamxTyler PTAL

[zz-jason]

/run-all-tests

[eurekaka]

LGTM

[winoros]

lgtm

[lysu]

do we need half-open stauts that plan/exec new & old at same time for compare and test?

[zz-jason]

yes, will be added once the new planner is functionally work.