[SPARK-55731][Streaming] Assign error class for multiple event time columns

[XdithyX]

What changes were proposed in this pull request?

This PR replaces _LEGACY_ERROR_TEMP_3077 with a proper error class, MULTIPLE_EVENT_TIME_COLUMNS(SPARK-55731).

The error is thrown when a stateful streaming operator receives an input DataFrame with more than one distinct event-time column while multiple event-time columns are not allowed.

This PR:

• Adds MULTIPLE_EVENT_TIME_COLUMNS to error-conditions.json
• Assigns SQLSTATE 42K09, matching nearby event-time analysis errors
• Updates both throw sites in WatermarkSupport
• Updates the existing EventTimeWatermarkSuite test to verify the new error condition

Why are the changes needed?

_LEGACY_ERROR_TEMP_3077 is a temporary legacy error class and should be replaced with a stable, descriptive error class.

The query behavior is already correct. This change improves the structured error reporting by assigning a real error condition for the multiple event-time columns case.

Does this PR introduce any user-facing change?

Yes.
For this error case, the error condition changes from _LEGACY_ERROR_TEMP_3077 to MULTIPLE_EVENT_TIME_COLUMNS.
The error message remains the same, and the query behavior is unchanged. The error now also has SQLSTATE 42K09.

How was this patch tested?

Tested by running
build/sbt 'sql / Test / testOnly org.apache.spark.sql.streaming.EventTimeWatermarkSuite -- -z "multiple event time columns in an input DataFrame for stateful operator is not allowed"'

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

Yes.

[XdithyX]

Hi @HeartSaVioR Can you please have a look? Thanks!

[XdithyX]

Hi @cloud-fan Can you please have a look? CI is now green.
Thank you