[GIE] Refine Partitioner trait to better support parallel scan

[BingqingLyu]

Is your feature request related to a problem? Please describe.

Refine Partitioner trait to better support parallel scan.

Currently, Partitioner define the get_worker_partitions function to specify the partition list that the worker (i.e., threads) is going to process:

 fn get_worker_partitions(&self, job_workers: usize, worker_id: u32) -> GraphProxyResult<Option<Vec<u64>>>;

However, we may have m workers and n partitions, where m>n. Thus, each worker need to scan only part of the partition. Take this case into consideration and refine the trait.