[SPARK-52184][SQL] Wrap external engine JDBC syntax errors with a unified exception

[alekjarmov]

What changes were proposed in this pull request?

This PR introduces a unified mechanism for handling SQL syntax errors thrown by external JDBC engines. It does so by:

• Adding a new error condition: JDBC_EXTERNAL_ENGINE_SYNTAX_ERROR.
• Introducing a utility method in JdbcUtils to catch and rethrow SQLException with a more informative SparkException if the dialect recognizes it as a syntax error.
• Extending multiple JdbcDialect implementations to provide best-effort syntax error detection via isSyntaxErrorBestEffort.
• Updating integration test suites to include a new shared test that verifies the wrapping behavior.

Why are the changes needed?

Current behavior rethrows the error if it's a syntax error on the external engine. Having it wrapped in a Spark exception enables for easier data scince.

Does this PR introduce any user-facing change?

The users would get a better error message.

How was this patch tested?

Adding new integration tests.

Was this patch authored or co-authored using generative AI tooling?

Coauthored with GitHub Copilot.

[MaxGekk]

As I can see the source is limited set of strings:

• "output schema resolution"
• "query execution"

How about to make 2 sub-conditions of "JDBC_EXTERNAL_ENGINE_SYNTAX_ERROR" instead of the parameter? In this way, we could:

1. avoid error text in source code. So, tech editors could modify the error-conditions.json w/o modifying/recompiling spark
2. in the future we might have errors in different languages.