Incorrect set semantics on the output of a correlated subquery

[scsmithr]

Describe the bug

Outputs of the correlated query aren't DISTINCTed, which may result in duplicated rows.

See TODO here: https://github.com/glaredb/glaredb/blob/888b7f39b04fb647dc8131514d6886ec35e04d81/crates/glaredb_core/src/execution/planner/plan_magic_scan.rs#L46

To Reproduce

This should only output two rows:

glaredb> select a, (select a + 1) from (values (1), (1)) v(a);
┌───────┬──────────┐
│ a     │ ?column? │
│ Int32 │ Int32    │
├───────┼──────────┤
│     1 │        2 │
│     1 │        2 │
│     1 │        2 │
│     1 │        2 │
└───────┴──────────┘

When using distinct values as the input, we produce the correct output:

glaredb> select a, (select a + 1) from (values (1), (2)) v(a);
┌───────┬──────────┐
│ a     │ ?column? │
│ Int32 │ Int32    │
├───────┼──────────┤
│     1 │        2 │
│     2 │        3 │
└───────┴──────────┘

Expected behavior

Correct number of rows returned.