Implement experimental Rust backend to address performance issues w/ large number of instructions

[MarquessV]

In short, the Python based compatibility layer that pyQuil uses to bridge the gap between quil-py and its own API is slow for large number of instructions. This compatibility layer could be pushed into Rust, and we can axe the dependency of quil-py all together. This could offer significant performance and code quality gains.

See #1755 for an in-depth writeup, and illustration of the benefits.