Benchmark VampIR circuits against native circuits in various zk libraries

[lopeetall]

We want to show that a circuit written in VampIR can be compiled to various backends and that VampIR circuits aren't too much worse than the native gadgets provided in those libraries.

If there isn't a matching native circuit available in the ZK library we will need to write a new VampIR circuit to match.

Benchmarks to consider:

1. Compilation time (VampIR -> backend circuit form)
2. Circuit size
3. Proof generation time
4. Verification time
5. Proof size

ZK libraries to target:

1. ECC/Halo2
2. ZK-Garage/Plonk
3. Circom/Snarkyjs
4. Aztec/Barretenburg
5. EspressoSystems/Jellyfish