Benchmarks and profiling

[arlyon]

This library has been written with the goal of doing as few allocations as possible however there are probably plenty for performance / speed gains to be had still. As an idea for potential pitfalls:

• creating too many intermediate ObjectLit
• inefficient config loading (parse and allocate the json config per file)
• inefficient parsing
• inefficient match for finding commands (rust_phf?)

I would like to build out flamegraphs for the snapshot tests to determine larger-than-expected execution times as well as run benchmarks on those to track perf over time.