Canonical test outputs / performance diffs

[Quantumplation]

Working on some continuous integration improvements for the sundae contracts, and came across some small pain points that I figured it was worth discussing:

1. I couldn't figure out a super easy way, just from the terminal, to grab a failing test and its traces; I can't do grep -A 1 for example, because it might span multiple lines, or 0; It'd be nice if the format was conducive to extracting the outputs for things like CI (or perhaps, there was a JSON formatted output)
2. I really want to add something like performance diffs to our CI; i.e. the ability to aiken check --diff baseline.output and have it tell me, for each test run, how the mem/cpu changed, or something similar.

None of these are urgent / critical, and may have other solutions beyond what I outlined above, just doing a brain dump so y'all can ponder on it.

[Quantumplation]

As a side note, I got this working. If you want to include it or something similar in the default github action, let me know:

[KtorZ]

A JSON-formatted output -> yes. This is just plain common sense, and I've been leaning towards that already for all commands. Basically, if a terminal output isn't ANSI-capable, then we should fallback to structured-JSON. This way, if you simply redirect the CLI output to a file, or if you're part of a context that can't properly display unicode and colors, you get something workable.

[KtorZ]

When you say "you", you mean you, right? 😏

[Quantumplation]

it's the royal you

[rvcas]

@waalge had this brought up before and we had discussed some kind of json output for it. Should be straight forward.

[rvcas]

Maybe just a bench.someformat file and then the --diff flag could auto detect the bench file or have it overridden.

[rvcas]

maybe --bench flag instead of --diff