Reduce sqlparse only test cases

[reata]

With v1.4.0, we introduce two parser implementations, namely sqlparse and sqlfluff. We run dual test using both parser and make sure the lineage result is exactly the same. But there're exceptions.

This issue is just to document all the test cases where sqlparse is the only option. They fail because sqlfluff fails to parse them. We'd like to reduce this list first by submitting issues to sqlfluff to solve these edge case.

test case	dialect
test_create_using_serde	hive
test_copy_into_path	tsql
test_select_from_unnest_with_ordinality	athena

In long term, as a validating and dialect-awareness parser, we'd like to ramp up usage of sqlfluff, bring it to the same level of capability, then gradually add cases where sqlfluff is the only option until ultimately get sqlparse removed.

[reata]

test_copy_into_path of tsql PR submitted: sqlfluff/sqlfluff#4985

[reata]

test_create_using_serde of hive is actually parsable with sqlfluff. We had character escaping issue in the test sql previously. This is now fixed via #413

[reata]

test_select_from_unnest_with_ordinality of athena PR submitted: sqlfluff/sqlfluff#4991

[reata]

All PR merged and released via https://github.com/sqlfluff/sqlfluff/releases/tag/2.1.4 and https://github.com/sqlfluff/sqlfluff/releases/tag/2.1.3.

We should be able to solve this by upgrading to sqlfluff v2.1.4