disk table supports duplicate key?

**Describe the feature you'd like**

One big difference between disk table and mem table is that disk table only allows one unique key per index.
For example, if the index is . the rows we inserted are:

|id |col1 | col2|
|-------| ------ | ----- |
1 | k1 | 1
2 | k1 | 1
3 | k2 | 1
4 | k2 | 2

In mem table, it will be:
|id |col1 | col2|
|-------| ------ | ----- |
1 | k1 | 1
2 | k1 | 1
3 | k2 | 1
4 | k2 | 2

However, in disk table, it will be:
|id |col1 | col2|
|-------| ------ | ----- |
1 | k1 | 1
3 | k2 | 1
4 | k2 | 2

because row 1 and row 2 have the same value of .

We should discuss whether we should support the same semantics for both disk table and memory table.

**Additional context**
One possible solution is to add a sequence number to the combined key for disk table.

Look at the disk table and memory table implementations to understand how keys are handled. The issue suggests adding a sequence number to the combined key for disk tables to allow duplicates. Start by examining the index structures and insertion logic, then design a change that maintains performance while supporting duplicate keys. Testing will involve verifying that duplicate rows are preserved in disk tables as they are in memory tables.

由索引模型根据 Issue 内容生成。