Remove Spark 3.2 compatibility code

[QCLyu]

Summary

• Remove all lteSpark32 branches from shared code (dead code since Spark 3.2 is no longer supported)
• Clean up SparkVersionUtil.scala (remove lteSpark32, lteSpark33)
• Delete orphaned shims/spark32/ files and Spark 3.2-specific test branches
• Rename KnownRowTypeForSpark33OrLater to KnownRowTypeWithDefault

Test plan

• Build with Spark 3.3, 3.4, 3.5 profiles to verify no compilation errors
• Run existing unit tests (no behavioral changes, only dead code removal)

Partially assisted by Claude Opus 4.6.

Closes #11379

[github-actions]

Run Gluten Clickhouse CI on x86

[QCLyu]

Hi @PHILO-HE , just thought carefully about your comment in the previous PR: #11495 (comment) This time I renamed KnownRowTypeForSpark33OrLater to KnownRowTypeWithDefault, instead of completely merging it with KnownRowType.

My rationale is "this: SparkPlan =>" created a self-type constraint: If you put this on KnownRowType directly, then any implementor of KnownRowType would have to be a SparkPlan. 

Please correct me if I'm wrong.

[github-actions]

Run Gluten Clickhouse CI on x86

[PHILO-HE]

Hi @PHILO-HE , just thought carefully about your comment in the previous PR: #11495 (comment) This time I renamed KnownRowTypeForSpark33OrLater to KnownRowTypeWithDefault, instead of completely merging it with KnownRowType.

My rationale is "this: SparkPlan =>" created a self-type constraint: If you put this on KnownRowType directly, then any implementor of KnownRowType would have to be a SparkPlan. 

Please correct me if I'm wrong.

@QCLyu, the constraint is acceptable maybe. This trait was introduced to categorize Spark Plans based on the data format. So I would advise merging them. Thanks.