[Coral-Spark] Modify coral hive parser and coral spark writer dialect to generate spark compliant escaped string literal

[rzhang10]

What changes are proposed in this pull request, and why are they necessary?

Without this patch, a string literal of select 'I'm' will be re-written in spark-sql output of select I''m, the '' double single quote escaping is a ANSI sql escaping standard but is not what is recognized by the spark sql parser. Instead, spark uses the backslash for escaping quotes inside quotes in sql string, i.e. the correct output should be select 'I\'m'.

This patch adds the customizations in the SparkSqlDialect to make the generated output sql consistency use single quote for quoting a outermost level string literal and use \' to quote internal literals occurences of ', for ", it will appear as-is since it will never need to be escaped because the outermost quote is always single quote.

How was this patch tested?

Added 6 unit tests to demonstrate the new behavior:

hive	spark
select 'abc'	select 'abc'
select "abc"	select 'abc'
select 'abc[\'xyz\']'	select 'abc[\'xyz\']'
select "abc['xyz']"	select 'abc[\'xyz\']'
select 'abc["xyz"]'	select 'abc[\'xyz\']'
select "abc[\"xyz\"]"	select 'abc["xyz"]'

Run regression tests against all views and results looks fine. Also tested against the view sql which exposed this issue on spark, the spark read result literals are correct now.

[ljfgem]

Thanks Raymond for this pr, could you add regression tests for both spark and trino? And could you also test it on spark shell against the target view to ensure it works as expected?

[rzhang10]

Thanks Raymond for this pr, could you add regression tests for both spark and trino? And could you also test it on spark shell against the target view to ensure it works as expected?

Yeah I just triggered int-test for both trino and spark, let's wait for the restults.

[ljfgem]

Thank you @rzhang10 for this PR, left one comment

[ljfgem]

LGTM, thank you @rzhang10 for the quick fix!