[core][spark] Support history-preserving replace table

[Zouxxyy]

Purpose

Support Spark replace-table flows for Paimon tables while preserving table history where possible.

This change adds staged replace support for REPLACE TABLE and CREATE OR REPLACE TABLE ... AS SELECT, using truncate plus schema replacement for compatible FileStore tables so old snapshots remain available for time travel. It also falls back to drop-and-create when the source or target is not a Paimon table, handles SparkGenericCatalog provider checks, and documents the Spark SQL behavior.

Follow-up TODOs:

• Support replace-table flows with RESTCatalog.
• Support replacing a table with different partitioning. This is tied to the underlying truncate mechanism and likely needs a separate PR.

Tests

Added coverage for catalog-level replace table, Spark SQL replace table, create-or-replace-as-select, incompatible schema replacement, table existence errors, provider/table type fallback, and time-travel reads after replace.

[JingsongLi]

We can implement REST Catalog in this PR too. This way we can have a global perspective.