Store computed metrics in BenchmarkResult and move report/pretty-print functionality into BenchmarkResult

[Bryysen]

Also added a total-time metric which represents how long the entire benchmark (all runs included) took to run. This is displayed when reporting/pretty printing. Also notably; we no longer print the result of a run after completion, the user must explicitly call prettyPrint on the BenchmarkResults instance. Updated examples to reflect this.

[Bryysen]

This can be considered (1/3) of the run overhaul. Note we deviate from https://pkg.go.dev/testing#hdr-Benchmarks by no longer pretty-printing/reporting right after a benchmark is complete. There are two reasons for why I think we should do that:

1. If a user runs the benchmark in a multithreaded context and each thread starts writing to stdout we can get race-conditions (we are not locking before writing to stdout). With this change they can now let the threads complete and join, and print the results from the main thread.
2. In the future more parsing options may be added, and the user might not want to print to stdout at all.

[hendriknielaender]

Excellent progress. The printing logic is quite lengthy and involves multiple steps. It might be beneficial to break this down into smaller, reusable functions for improved readability and maintainability. What do you think?

[Bryysen]

Excellent progress. The printing logic is quite lengthy and involves multiple steps. It might be beneficial to break this down into smaller, reusable functions for improved readability and maintainability. What do you think?

I agree with this. In the future we may add functionality for the user to choose which metrics to display, having split everything up into individual functions makes that easier. I'll address the feedback as soon as I have some time.

[Bryysen]

Note: In 3313c40 I make use of BenchmarkResults.getColor the way I imagine it was initially intended to be used, however I haven't touched on the function internally nor have I checked that it does what we want it to when printing multiple results. In general we need more tests but I'm putting that off for now since there are a lot of changes in the works.

Edit: By the way if you don't agree with letting the caller select the writer let me know and I'll revert to using stdout directly in the pretty-print functions.

[hendriknielaender]

Just a couple of small things, but really cool stuff. 👍