feat!: Update to ACVM v0.11.0

[phated]

This reworks the backend to leverage the changes for fallible functions. It introduces an error enum with admittedly bad variant names and uses them to map error cases throughout the codebase so panic and/or unwrap isn't used to handle most values.

There are still some TODOs to cleanup even more unwrap cases and it relies on something like noir-lang/acvm#251, but this should be a solid skeleton for how a backend can surface errors to nargo.

Closes #159

[kevaundray]

You can change the commit to point to acvm 0.11 -- Is this ready for review? If not, then I can push a new PR with the fallible traits but has the same behaviour as without the fallible traits.

Main goal here is to get Keccak upstreamed -- Alternatively, I can add Keccak into the pwg for now by copying this code into here.

Then removing it once we use acvm > 0.11

[phated]

Promoting from draft because acvm 0.11 was released but we still need to workshop the errors and stuff.

[phated]

I've completed my TryInto TODOs and split the wasm errors into their own enum. src/merkle.rs still has a lot of unwraps but I don't think we need to update that for this to land.

Besides error variant naming and error messages, I think this should be good-to-go. Will wait on @TomAFrench to do a full review and give feedback on those.

[TomAFrench]

A couple of these error types don't need to exist in my mind (comments below).

We've also bundled in a change which changes WASMValue to cover not just return values but all arguments as well. I'm going to split off this change so I can review errors in isolation.

[TomAFrench]

I think that we can remove a number of errors which are included here. I don't think we should expose these as users never need to be concerned with them. I don't think it's worth blocking over however as we don't have semver set up yet so every release is breaking and we can just remove them progressively.

.

[TomAFrench]

Some thoughts on error variants

// These variants shouldn't exist as we can prevent them at compile time
SchnorrConvert(Vec<u8>),
SchnorrSlice { source: TryFromSliceError },
FieldElementSlice { source: TryFromSliceError },
FieldToArray(usize, usize),

// Can be replaced with a MalformedOpcode(Opcode) error
// This error can be defined down in ACVM and be shared across backends
MissingOutput(usize),
MissingLeaf,
MissingLeafIndex,
MissingPublicKeyX,
MissingPublicKeyY,
MissingSignature(usize),
MissingPublicKey(&'static str, usize),

// I think we can leave this as a panic tbh.
// Calling a backend with an unsupported opcode is calling the `solve` function outside of contract.
Aes,

// Unsure what to do these
Pow2CeilOverflow(u32),
WasmError,

[phated]

// These variants shouldn't exist as we can prevent them at compile time
SchnorrConvert(Vec<u8>),
SchnorrSlice { source: TryFromSliceError },
FieldElementSlice { source: TryFromSliceError },
FieldToArray(usize, usize),

I looked through the compiler and we currently aren't preventing them at compile time so we should leave them for this to land sooner. We need to fix things in bb-sys to ensure this at compile time.

// Can be replaced with a MalformedOpcode(Opcode) error
// This error can be defined down in ACVM and be shared across backends
MissingOutput(usize),
MissingLeaf,
MissingLeafIndex,
MissingPublicKeyX,
MissingPublicKeyY,
MissingSignature(usize),
MissingPublicKey(&'static str, usize),

I'm fine joining these into one error, but these are the TryFrom<Circuit> and shouldn't be defined in ACVM

// I think we can leave this as a panic tbh.
// Calling a backend with an unsupported opcode is calling the `solve` function outside of contract.
Aes,

I'm going to leave this because it's easier to not have to grep for all the different types of ways to panic.

// Unsure what to do these
Pow2CeilOverflow(u32),
WasmError,

We can leave them as-is then.

[phated]

I think this PR is good-to-go and we should hold any other changes/PRs until it lands. We need to get this stuff upstreamed.

[kobyhallx]

This is great! Thank You!