Fix redundant data copying in unnest

[dmitriibugakov]

Which issue does this PR close?

Closes #13237.

Rationale for this change

In the current implementation, during the process where we align all columns to match the length of the longest unnest operation, it can result in excessive data copying and lead to unexpected errors for queries such as:

select * from (select unnest(range(0, 100000)) id) t inner join (select unnest(range(0, 100000)) id) t1 on t.id = t1.id;

This can trigger errors like attempt to add with overflow

What changes are included in this PR?

Introduced repeat_mask to determine which columns need to be repeated or replaced with nulls based on:

1. Whether the column is required in future unnesting levels.
2. Whether the column is involved in unnesting at all.

Ensures only necessary columns are repeated, reducing unnecessary data copying.

Are these changes tested?

Yes

Are there any user-facing changes?

No

[alamb]

Thanks @demetribu -- this looks very cool

@duongcongtoai , as the author of #12836 do you think you might be able to review this PR as well?

[duongcongtoai]

nice, i will take a look

[duongcongtoai]

LGTM

[alamb]

Thanks @demetribu and @duongcongtoai ❤️

[dmitriibugakov]

Thanks for review @alamb and @duongcongtoai