[Feature] concurrent write of struct sinker

[caiq1nyu]

Motivations

the current structure's write model only supports serial writes（serial_parallelizer.rs）, which can lead to performance bottlenecks when there are many tables.

[extractor]
extract_type=struct

[sinker]
sink_type=struct

[parallelizer]
parallel_type=serial

direct and rough concurrency may cause data problems, such as creating an index first but the table does not exist

Solution

• provides a new parallelizer implementation that includes data conflict management in concurrent scenarios.
• mark conflict information in advance in the data fetched in the extractor (xx-struct-extractor.rs)

'tables' depend on 'database';
'indexs' depend on 'tables';
'tables have FK' deplend on 'other tables' and so on