Concurrently written rows in REORGANIZE PARTITION may not be cleaned up

[mjonss]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

Uncomment these lines, make failpoint-enable and run the test TestMultiSchemaReorganizePartition

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

Test to pass, no old entries in old partitions ranges

3. What did you see instead (Required)

Reorganized (old partition) had still entries even after GC!

[2024/11/20 01:47:52.645 +01:00] [INFO] [multi_domain_test.go:704] [HaveEntriesForTableIndex] [tableID=111] [indexID=0] [count=1]
    multi_domain_test.go:472: 
        	Error Trace:	/Users/mattias/repos/tidb/pkg/ddl/tests/partition/multi_domain_test.go:472
        	            				/Users/mattias/repos/tidb/pkg/ddl/tests/partition/multi_domain_test.go:669
        	            				/Users/mattias/repos/tidb/pkg/ddl/tests/partition/multi_domain_test.go:475
        	Error:      	Should be false
        	Test:       	TestMultiSchemaReorganizePartition
        	Messages:   	Reorganized partition id 111 for table id 110 has still entries!
--- FAIL: TestMultiSchemaReorganizePartition (0.36s)

4. What is your TiDB version? (Required)

tidb_version(): Release Version: v8.5.0-alpha-163-g50b5cd27d4
Edition: Community
Git Commit Hash: 50b5cd27d41309a9b83d9c6e3529e15dd2cd60f9
Git Branch: master
UTC Build Time: 2024-11-20 00:54:03
GoVersion: go1.23.2
Race Enabled: false
Check Table Before Drop: false
Store: unistore