[core][spark] Fix RTAS reads after incompatible nested type replace

[Zouxxyy]

Purpose

Fix CREATE OR REPLACE TABLE ... AS SELECT after an incompatible nested type replace.

The issue happens in the two-snapshot RTAS flow:

1. a truncate snapshot clears the old visible files
2. the following RTAS append snapshot writes the new result

Without extra handling, the RTAS append snapshot can still inherit stale snapshot state from the truncate snapshot. This is especially problematic for incompatible schema replacement, where latest-table reads may observe old metadata/state that should not survive the replace.

This change:

• adds a Spark SQL regression in DDLTestBase that reproduces the incompatible nested-type RTAS read failure
• resets inherited manifest state for the RTAS append-after-truncate path
• also clears inherited index manifest and statistics metadata in that path
• adds a focused core unit test to verify the index/statistics reset behavior

Tests

• mvn -pl paimon-core -am -Pfast-build -DfailIfNoTests=false -Dtest=FileStoreCommitTest#testRtasAppendAfterTruncateResetsInheritedIndexAndStats test
• mvn -pl paimon-spark/paimon-spark-3.5 -Pspark3,fast-build -DfailIfNoTests=false -DwildcardSuites=org.apache.paimon.spark.sql.DDLTest -Dtest=none test
• mvn -Pspark3 spotless:apply

[JingsongLi]

+1