Chore!: fix mypy issues, use _parse_table_parts instead of _parse_table

[georgesittas]

This will resolve mypy issues created due to the latest sqlglot changes.

When we bump sqlglot to v14.1.0, I'll bump its version here too and the CI will be fixed.

[georgesittas]

@tobymao can you take a look here? Seems like this became invalid due to tobymao/sqlglot@fbf5f47.

Note that by assigning a list of Columns we'll generate a different AST here than what parse_one outputs:

>>> import sqlglot
>>> sqlglot.parse_one("CREATE INDEX my_idx ON tbl (a, b)")
(CREATE this:
  (INDEX this:
    (IDENTIFIER this: my_idx, quoted: False), table:
    (TABLE this:
      (IDENTIFIER this: tbl, quoted: False)), columns:
    (ORDERED this:
      (COLUMN this:
        (IDENTIFIER this: a, quoted: False)), desc: False, nulls_first: True),
    (ORDERED this:
      (COLUMN this:
        (IDENTIFIER this: b, quoted: False)), desc: False, nulls_first: True)), kind: INDEX)

[tobymao]

yea that was wrong

[tobymao]

columns is fine

[georgesittas]

Cool, thanks for verifying. Doing a final pass on the codebase to see if there's any other place where we could have a regression, but it's probably good to go.