fix(ir): fix columns size for vector trace access

[tohrnii]

Addressing #289

[grjte]

Nice catch, thank you! This looks correct to me. However, let's add a test for the issue that you're fixing before merging this.

[tohrnii]

let's add a test for the issue

I've modified the evaluators integration test.

[grjte]

Thank you - one more request, sorry!

[grjte]

To really test both cases effectively, I think we should pass a column group to an evaluator function, e.g.

ev are_all_binary([c[3]]):
    enf is_binary([c[i]]) for i in 0..3

I know this is a stupid example, but the point is just to test passing in a SymbolType with AccessType::Default where the size is > 1.

I think this is my oversight from before the previous fix, since I think it was captured in another test elsewhere instead, but I'd appreciate it if you could update it

[tohrnii]

Actually we don't yet support variables as indices for comprehensions in the parser. Maybe we should prioritize this issue since a lot of our constraints depend on this. Also, we haven't merged constraint comprehension for evaluators yet so I've added test for:

ev are_all_binary([c[3]]):
    enf c^2 = c for c in c

[grjte]

Ah ok good point, thanks. That works! That's not the part I cared about for the test anyway

[grjte]

Thank you, looks good!