feat: introduce new benchmarking script

[ielashi]

Problem

We've been relying on criterion as our way of running benchmarks and getting reports. However, criterion has been falling short for a number of reasons:

1. Benchmarks can only have one measurement. In our case, the measurement was "instructions", but that's not good enough. For our use-case, we'd like to have multiple measurements per benchmark (e.g. instructions, stable memory pages, etc.).
2. criterion always compares the results to the previous run. In almost all cases, I'm interested in measuring the difference in performance relative to the last commit, and there isn't a good of enforcing that workflow in criteron.

Solution

I decided to spend a few hours to replace our reliance on criterion altogether. We now have our very own benchmarking script that runs the benchmarks. A benchmark can have as many measurements as we want, and it neatly persists these benchmarks to our repo along with a report on how the benchmarks have changed.

Here's an example of the report that it prints out to the terminal:

Notes:

• As you can tell from the screen shot above, I added the "stable_memory_size" as a new measurement.
• I had to bump up the rust version as clap needed a more rust >= 1.70.