[BUG]索引扫描删除记录不能完全删除

Describe the bug
走索引扫描删除记录时删除不能全部删除

Environment

• OS Version: ubuntu22.04
• CPU Arch(x86/arm): x86
• Compiler: GCC 11.4.0 x86_64-linux-gnu
• Others:

Fast Reproduce Steps(Required)
Steps to reproduce the behavior:
miniob> create table t (id int,age int);
SUCCESS
miniob> create index t_id_idx on t (id);
SUCCESS
miniob> insert into t values (1,10);
SUCCESS
miniob > insert into t values (1,11);
SUCCESS
miniob > insert into t values (1,12);
SUCCESS
miniob > select from t;
id|age
1|10
1|11
1|12
miniob > delete from t where id 1;
SUCCESS
Expected behavior
记录应该全部被删除
miniob > select from t;
id|age
Actual Behavior
What is the result? picture is allowed
中间的记录被跳过没有被删除
miniob > select from t;
id|age
1|11

Additional context

• 导致该问题的原因是在RC BplusTreeScanner::next_entry(RID& rid, bool idx_need_increase)这个函数中查找下一个entry时，会把iter_index_++，但删除一条记录会将其对应的索引向前覆盖memmove(__item_at(index), __item_at(index + 1), (size() - index - 1) * item_size())，iter_index_++就直接跳过了下一条要获取的记录了。

Start with BplusTreeScanner::next_entry(RID& rid, bool idx_need_increase) and reproduce the shown duplicate-key DELETE scenario in the miniob prompt. Trace how the scanner advances while an index entry is removed, then verify that deleting all rows matching id 1 leaves no records in the final SELECT result.