feat: auto-collection of {mem,time}-profiling

[cyphersnake]

Right now we have time-profiling & mem-profiling, but it is collected manually. I propose to work out a methodology, what parameters we want to evaluate, choose targets and move towards them during execution.

We can summarize the results and e.g. attach them to CI at some stage (e.g. at each PR, or at each main update, or at each intermediate release (git-tag) etc).

But first of all we need to understand what parameters we want to monitor.