fix(ssa refactor): recursive branch analysis

[joss-aztec]

Description

Problem*

Resolves GH-1664

The existing implementation of analyze_function used a queue + stack-based approach to to associating a branch start with an end. I couldn't convince myself that there was a way to make this approach safe, so I've written a recursion-based alternative that passes the failing case surfaced by GH-1664.

Summary*

This alternative recursion based-approach associates a branch start to its end at the point that a recursion exits, such that the association is encapsulated. A disadvantage to this approach is that it does result in extra allocations, because visited hash sets and Steppers are allocated for each level of recursion.

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.

[jfecher]

I went back and forth on this a little bit wondering if we could leverage an existing graph algorithm like Least Common Ancestor (with the edges reversed to find branch ends), but that wouldn't give use the list of in-between blocks. This change is also still fairly short so I think it is good to merge since it fixes the bug in question.