[SPARK-42921][SQL][TESTS] Split timestampNTZ/datetime-special.sql into w/ and w/o ansi suffix to pass sql analyzer test in ansi mode

[LuciferYang]

What changes were proposed in this pull request?

After #40496, run

SPARK_ANSI_SQL_MODE=true build/sbt "sql/testOnly org.apache.spark.sql.SQLQueryTestSuite"

There is one test faild with spark.sql.ansi.enabled = true

[info] - timestampNTZ/datetime-special.sql_analyzer_test *** FAILED *** (11 milliseconds)
[info]   timestampNTZ/datetime-special.sql_analyzer_test
[info]   Expected "...date(999999, 3, 18, [false) AS make_date(999999, 3, 18)#x, make_date(-1, 1, 28, fals]e) AS make_date(-1, ...", but got "...date(999999, 3, 18, [true) AS make_date(999999, 3, 18)#x, make_date(-1, 1, 28, tru]e) AS make_date(-1, ..." Result did not match for query #1
[info]   select make_date(999999, 3, 18), make_date(-1, 1, 28) (SQLQueryTestSuite.scala:777)
[info]   org.scalatest.exceptions.TestFailedException:

The failure reason is the last parameter of function MakeDate is failOnError: Boolean = SQLConf.get.ansiEnabled.

So this pr split timestampNTZ/datetime-special.sql into w/ and w/o ansi to mask this test difference.

Why are the changes needed?

Make SQLQueryTestSuite test pass with spark.sql.ansi.enabled = true.

Does this PR introduce any user-facing change?

No

How was this patch tested?

• Pass GitHub Actions
• Manual checked SPARK_ANSI_SQL_MODE=true build/sbt "sql/testOnly org.apache.spark.sql.SQLQueryTestSuite"

[LuciferYang]

cc @dtenedor @HyukjinKwon FYI

[dtenedor]

I am so sorry to break the build again :| thanks for fixing it! It looks like we need separate regular and ANSI test cases now!

[LuciferYang]

ignore in ThriftServerQueryTestSuite, same reason as timestampNTZ/datetime-special.sql

[HyukjinKwon]

Merged to master.

[LuciferYang]

Thanks @HyukjinKwon @dtenedor