Optimize the performance consumption of ShardingSphere and Seata to parse SQL repeatedly

## Feature Request

### Is your feature request related to a problem?

No.

### Describe the feature you would like.


At present, ShardingSphere integrates Seata to implement Base transactions, and has achieved good results. But there is still a problem, both ShardingSphere and Seata parse the SQL when processing it, which leads to additional performance overhead.

In order to optimize the performance of the integrated use of ShardingSphere and Seata, we need to integrate the SQL parsing of ShardingSphere and Seata, and pass the AST of ShardingSphere to Seata to avoid the problem of repeated SQL parsing.

Start by investigating the ShardingSphere and Seata integration described in the issue, focusing on where both systems parse SQL. Determine how ShardingSphere's AST could be passed to Seata and how performance improvements would be measured; the issue names no files or tests.

Written by the indexing model from the issue text.