[spark] Prevent table deletion for self referencing RTAS

[ArnavBalyan]

Purpose

• A self referencing replace table on an existing table fails the query and accidentally performs a data drop.
• This is because Spark drops the table before evaluating the query. The query always fails while reading the deleted manifest leading to original table being lost permanently.
• Add guard to ensure this operation does not happen for partitioned table rtas and we can avoid data loss.

Tests

• UT