anti semi join with other condition may get wrong results

[windtalker]

Bug Report

Please answer these questions before submitting your issue. Thanks!

During the join probe stage, in order to make sure that Intermediate block's size is not bigger than max_block_size, one probe block maybe probed multiple times, each time only [probe_process_info.start_row, probe_process_info.end_row) rows is processed. If the join has other condition, after probe the hash table, it need to call handleOtherConditions to further evaluate the other condition. In handleOtherConditions, the input argument anti_filter and offsets_to_replicate is expected to be a vector of size probe_process_info.end_row - probe_process_info.start_row, however, for anti-semi join, the caller only cut offsets_to_replicate to the correct size(https://github.com/pingcap/tiflash/blob/v7.5.0/dbms/src/Interpreters/Join.cpp#L1297-L1307), for filter, it is not handled, which will make handleOtherConditions return wrong result in some cases.

This bug only affect v6.6 ~v7.5

1. Minimal reproduce step (Required)

2. What did you expect to see? (Required)

3. What did you see instead (Required)

4. What is your TiFlash version? (Required)

[windtalker]

closed by #8792

[seiya-annie]

/found customer