Don't update next Key in cache lead to update the wrong next key to storage when modifying column interrupted

[zimulala]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

create table t1 (c1 int, c2 int, c3 int, index idx(c2), index idx1(c1, c2));
alter table t1 change c2 c2 mediumint;
admin check table t1;

When executing modify column statement in the write-reorg state(handling backfill of the first index), encounter a retried error like TxnRetryableError or ErrNotOwner(This error is currently being handled incorrectly).

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

Execute successfully.

3. What did you see instead (Required)

Data indexes are inconsistent.

At present, because ErrNotOwner does not have this problem, TxnRetryableError is difficult to appear (generally a high success rate, in addition, you can retry 10 times). So I set it to [severity/moderate].

4. What is your TiDB version? (Required)

Release Version: v6.5.0-alpha-271-g5127ad25b
Edition: Community
Git Commit Hash: 5127ad25b5e05ae550be148ffaa50c41b2348c80
Git Branch: master
UTC Build T