New benchmarks

[LDiazN]

This PR adds extra benchmarks to the project to have a better understanding of performance bottlenecks.

Also a flamegraph library is added to have a better understanding of the cost per function call.

Some of the results we found out the following avg times:

• verification time (measurement hashing + zkp + Python overhead): 34ms
• zkp: 13ms
• hashing (1mb worth of data): 1ms

Here's a screenshot of a generated flamegraph:

• It seems like the Python bindings are introducing a lot of overhead
• Most time is still spent on verification
• No Python-related operations are done during verification or hashing, we could potentially release the GIL during this time to allow true parallel execution (see https://pyo3.rs/v0.29.0/parallelism)

closes #46