Performance regression introduced by pr/31547

[tiancaiamao]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

See http://www.zenlife.tk:18081/

Verify it:

git checkout 9c64224c4951fdffb05f284b011c48c50c188a04
cd session;
go test -run XXX -bench BenchmarkInsertIntoSelect -benchmem -cpuprofile cpu2.out -memprofile mem2.out -benchtime 30s

Before this commit:

BenchmarkInsertIntoSelect-16    	     418	  85843376 ns/op	104122866 B/op	  405474 allocs/op

After this commit

BenchmarkInsertIntoSelect-16    	     264	 135987408 ns/op	151347408 B/op	  785351 allocs/op

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

3. What did you see instead (Required)

4. What is your TiDB version? (Required)

[tiancaiamao]

Ref #25754

[tiancaiamao]

Profile before

vs After

@ekexium

[tiancaiamao]

The other regression BenchmarkInsertWithIndex is also by this change.