WIP: Rough out the idea of reference tables for kwargs and situations

[Rowlando13]

No description provided.

[kdeldycke]

I really like the exhaustivity of the table. Because we can rely on it to track inconsistencies and as a truth table for tests. I think I will reuse that pattern to document the behavior of boolean flags in #3239

[Rowlando13]

@kdeldycke Thanks. After doing this, I realized there are just too many arguments to resolve in a good way, because there are just so many edge cases. Not sure exactly what do about it yet, but plan to detail it out in an issue.

[Rowlando13]

To resolve this will need to keep options and arguments public API but have a private API that breaks arguments and options down into more parts. To solve the many silent failures the current setup causes, will need to have a --check setting that can be run during tests to warn people.

[kdeldycke]

@Rowlando13 if this is becoming too overwhelming, starts by looking at the @parametrize cases of the unittests: they're the ground truth. And maybe a better way of understanding the general rules and the edge-cases.