[VL][Iceberg] Support MERGE INTO

[infvg]

Enable native scans and writes for Iceberg MERGE INTO in copy-on-write mode.

Iceberg uses SparkCopyOnWriteScan and file-path/row-position metadata for these operations. The native scan path previously did not support this scan type or handle the required metadata.

This PR:

• Supports SparkCopyOnWriteScan in Iceberg scan detection, table lookup, and scan-task extraction.
• Maps copy-on-write file-path and row-position columns to native metadata and row-index columns.
• Retrieves the underlying Iceberg table from the scan so validation works when row-level operations wrap the Spark table.
• Extends row-index detection to include Iceberg row-position columns and columns at schema index zero, while preserving Spark’s schema validation and backend capability checks.

[github-actions]

Run Gluten Clickhouse CI on x86

[github-actions]

Run Gluten Clickhouse CI on x86

[github-actions]

Run Gluten Clickhouse CI on x86

[github-actions]

Run Gluten Clickhouse CI on x86

[zhouyuan]

Cc: @wangyum

[wangyum]

Will test and verify it next week.

[infvg]

Hi @wangyum, could you please take a look whenever you can?

[jinchengchenghh]

Please write the PR description

[github-actions]

Run Gluten Clickhouse CI on x86