Break up and/or build JSON scenarios from code

[notatallshaw]

I'm finding reading through the JSONs a little challenging, I have two suggestions (and would be happy to work on them at some point):

• Break them down into more granular scenarios, e.g. I would split prerelease into "prerelease-direct", and "prerelease-transitive"
• Build JSONs from a code model, allowing to easily encode a number of very similiar tests that only vary by small variables, e.g. ">" and ">=" should pass, or a sequence of tuples of succcess/failures [(">", true), (">=", true), ("==", true), ("!=", false), ("<=", false), ("<=", false)] rathar than writing out 6 different scenarios by hand

[zanieb]

I don't feel strongly about reducing the size of the files, i.e. we could have scenarios/prerelease/<kind>.json — I'm a little unsure what this helps with though you just end up needing to navigate across more files?

I think we may want to support JSON5 or YAML to allow comments in the files, that seems helpful for readability.

I don't see any problem with generating scenarios from code, but I think that would be a step before packse which is intended to build packages from scenarios. I don't think packse should be complicated by reading code to generate scenarios, but generating scenarios from code and then processing them with packse sounds nice.