refactor: change some hashbrown RawTable uses to HashTable (round 2)

[crepererum]

Which issue does this PR close?

For #13433, but only parts of it.

Rationale for this change

Prepare hashbrown 0.15 upgrade.

What changes are included in this PR?

• HashTableAllocExt
• a few migrations

Are these changes tested?

Existing tests pass.

Are there any user-facing changes?

No.

[crepererum]

Picking this one, but this pattern occurs multiple times:

That SAFETY statement is just plain wrong: while iterating over the map, we erase elements from it. That is THE prime example for "concurrent modification of containers" and why Rust's lifetimes/reference system prevents that. I'm kinda surprised that this hasn't exploded yet.

[crepererum]

The test failure (full logs here):

2024-11-22T12:00:01.8774000Z External error: query result mismatch:
2024-11-22T12:00:01.8776864Z [SQL] select make_array(make_array(1)) x UNION ALL SELECT make_array(arrow_cast(make_array(-1), 'LargeList(Int8)')) x;
2024-11-22T12:00:01.8778565Z [Diff] (-expected|+actual)
2024-11-22T12:00:01.8778975Z -   [[-1]]
2024-11-22T12:00:01.8779248Z -   [[1]]
2024-11-22T12:00:01.8779488Z +   [[1]]
2024-11-22T12:00:01.8779735Z +   [[-1]]
2024-11-22T12:00:01.8780003Z at test_files/union.slt:775

Is the order of UNION ALL even defined?

[Dandandan]

The test failure (full logs here):

2024-11-22T12:00:01.8774000Z External error: query result mismatch:
2024-11-22T12:00:01.8776864Z [SQL] select make_array(make_array(1)) x UNION ALL SELECT make_array(arrow_cast(make_array(-1), 'LargeList(Int8)')) x;
2024-11-22T12:00:01.8778565Z [Diff] (-expected|+actual)
2024-11-22T12:00:01.8778975Z -   [[-1]]
2024-11-22T12:00:01.8779248Z -   [[1]]
2024-11-22T12:00:01.8779488Z +   [[1]]
2024-11-22T12:00:01.8779735Z +   [[-1]]
2024-11-22T12:00:01.8780003Z at test_files/union.slt:775

Is the order of UNION ALL even defined?

It isn't / shouldn't be AFAIK

[Dandandan]

Test was fixed in #13547

[Dandandan]

Changes look good, can we run some benchmark.