feat(experimental): Compile match expressions

[jfecher]

Description

Problem*

Working towards #988

Summary*

Adds support for compiling match expressions. Specifically:

• During elaboration, convert match expressions to decision trees where:
  • No tag value check is repeated
  • Nested enums, tuples, and integer types are supported
  • NO errors or warnings are issued, they're all stubbed out for future PRs (just write perfect code)
  • Match cases with identical branches are deduplicated.
    • Duplicated branches can happen quite often and we could be more aggressive with this in the future since nested matches won't be deduplicated across nested cases currently.
  • The match compiler supports more pattern types than we currently parse. In addition to basic constructors & variable patterns it also supports:
    • Range patterns for ranges of integers
    • Pattern guards (_ if .. => ...)
    • We do not support matching on strings, arrays, or slices but could consider this for future work.
• Support for lowering decision trees into SSA via many jmpifs
  • We'll want to add a TerminatorInstruction::Switch eventually but I left this for future work

So you can now execute & prove code containing match expressions.

Additional Context

Quite a few parts of match expressions are still left as TODOs for future PRs to keep this PR "small":

• Redundancy checking
• Missing case analysis
• Errors in general (e.g. error: type '&mut T' can't be matched on)
• There is an odd stack overflow when using variants which reference themselves in their generics. E.g. MyOption<MyOption<Field>> will cause a stack overflow in Monomorphizer::convert_type.
• We could deduplicate cases more aggressively (check all cases, not just neighboring cases, and check nested cases not just cases of the current sub-match).
• Can trigger an SSA panic when multiple match blocks all jump to the same return block during flattening. It wasn't possible for > 2 blocks to jump to the same block directly before so this triggers an assert. Flattening has difficulty flattening matches since it is used to if branches with only 2 cases. Match expressions panic during SSA flattening if they're on a runtime value with > 2 branches #7323

Documentation*

Check one:

• No documentation needed.
• Documentation included in this PR.
• [For Experimental Features] Documentation to be submitted in a separate PR.

PR Checklist*

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

[vezenovm]

Approving, just have those few comments for tracking follow-ups

[github-actions]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Test Suite Duration'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: 0466e06	Previous: f37eedc	Ratio
AztecProtocol_aztec-packages_noir-projects_noir-protocol-circuits_crates_reset-kernel-lib	11 s	9 s	1.22

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench