[Query] Retry schema queries after transient internal write failures

[Caideyipi]

Description

Problem

A table-model SELECT may execute an internal TableNodeLocationAddNode write before reading a SchemaRegion. When that write exhausts the Ratis-level retry during a transient leader step-down, the retryable write failure was returned directly to the client as a statement execution error.

Fix

• Reuse the existing write retry policy for this internal schema-query write.
• Convert retryable internal write failures to DISPATCH_ERROR and set readNeedRetry, allowing the query scheduler to retry the query.
• Preserve non-retryable failures and their original status.

Test

mvn -pl iotdb-core/datanode -am clean -Dtest=TableSchemaQueryWriteVisitorTest test

This PR has:

• been self-reviewed.
• added comments explaining the intent of the retry conversion.
• added unit tests covering retryable and non-retryable internal write failures.

Key changed/added classes

• TableSchemaQueryWriteVisitor
• TableSchemaQueryWriteVisitorTest