Add optimization passes for native LastJoin

Native LastJoin is implemented in OpenMLDB Spark distribution and make great performance improved. We should add more optimization passes like LeftOuterJoin to leverage the improvement of Spark.

Look at the existing native LastJoin implementation in the OpenMLDB Spark distribution to understand its optimization passes. Investigate how LeftOuterJoin and similar joins are currently handled. Determine what changes are needed to extend the optimization passes, likely in the Spark-related codebase within the project. Testing will involve verifying performance improvements for the new optimization passes.

Written by the indexing model from the issue text.