[Improvement] The execution time of spark connector is 4 times that of native spark3.3.2 when running tpcds sql99

### What would you like to be improved?

A comparative test using tpcds 1000G data found that the execution time of spark connector is 4 times that of native spark 3.3.2.
Comparing the physical execution plans of the two, the Gravitino Spark Connector does not use the predicate pushdown and dynamic partition pruning optimization strategies.

[spark-gravitino-tpcds性能对比测试.xlsx](https://github.com/user-attachments/files/19861742/spark-gravitino-tpcds.xlsx)
[tpcds-query1-gravitino执行计划.txt](https://github.com/user-attachments/files/19861741/tpcds-query1-gravitino.txt)
[tpcds-query1-spark执行计划.txt](https://github.com/user-attachments/files/19861740/tpcds-query1-spark.txt)

### How should we improve?

_No response_

Start by comparing tpcds-query1-gravitino执行计划.txt with tpcds-query1-spark执行计划.txt and reviewing the attached TPCDS comparison workbook. Identify where predicate pushdown and dynamic partition pruning are absent from the Gravitino Spark Connector plan; done means both optimizations are used and the execution-time gap is remeasured.

Written by the indexing model from the issue text.