[CORE][VL] V2 columnar write

The topic is to enable DSV2 write in Gluten, also in preparation for Iceberg support.

This will involve:

1. Have a V2 columnar write API
2. Test: Add columnar version of `InMemoryTableCatalog` (e.g., `ColumnarInMemoryTableCatalog`) that fits into V2 columnar write API, but not necessarily do columnar processing internally (could reuse `InMemoryTableCatalog`'s row-based storage after the data passes through the columnar API)
3. Enable `DataFrameWriterV2Suite`

Start by locating the V2 write API and DataFrameWriterV2Suite, then inspect InMemoryTableCatalog to determine how a columnar test catalog would fit. Done means providing a V2 columnar write API, adding the columnar InMemoryTableCatalog test coverage, and enabling DataFrameWriterV2Suite.

Written by the indexing model from the issue text.