fix(ssa refactor): Ignore array out of bounds errors when enable_side_effects is false

[jfecher]

Description

Problem*

When enable_side_effects is false, we are meant to be in a false branch of an if that was not taken, so we should ignore side effects causing the program to fail, such as array out of bounds errors. I've added a check for this to only panic for array index of out of bounds errors if enable side effects is not known to be false.

This fixes the bug in #1629 but means the issue will still appear if the if uses a witness as its condition, as it would not be known to be false. I've changed the logic to only panic if enable_side_effects is known to be true rather than not known to be false.

Resolves #1629

Summary*

Documentation

• This PR requires documentation updates when merged.

  • I will submit a noir-lang/docs PR.
  • I will request for and support Dev Rel's help in documenting this PR.

Additional Context

PR Checklist*

• I have tested the changes locally.
• I have formatted the changes with Prettier and/or cargo fmt on default settings.

[ludamad]

Hmm interesting problem, longer term we may have to ignore even more array out of bounds I guess you're implying? LGTM

[ludamad]

Cool that change makes sense to me

[ludamad]

Maybe a comment about the logic that led here? Probably surprising that it's usually ignored at first read

[jfecher]

detecting if both sides would have an error, possibly a conservative setting that doesn't allow an error on either side but that I'm more speculative on

What do you mean by both sides here?

[ludamad]

Deleted that comment, thought this came up more generally