Use a dedicated TestResult object instead of keeping ConjectureData around

[DRMacIver]

The main motivation for doing this was memory usage, but I think it actually makes the API much cleaner to have a logical distinction between its data and its result, and it's something I've vaguely felt I should do for a while but never got around to.

[Zalathar]

Is #411 still relevant here, or is it fine for this type to be named TestResult?

[Zac-HD]

Is #411 still relevant here, or is it fine for this type to be named TestResult?

If it's something we should avoid doing, it would be good to add a linter for it (in another PR). flake8-alfred looks like it can only scan for exact matches, but I think git grep "def test|class Test" -- hypothesis-python/src/hypothesis is probably enough to handle this case!

The specific reason to avoid this no longer applies to modern versions of Pytest, but it's a good general principle and we do support and test against older versions where it would matter.

[DRMacIver]

Is #411 still relevant here, or is it fine for this type to be named TestResult?

Oh yeah, that's a good shout. It's not fine and I'll change the name.

[DRMacIver]

If it's something we should avoid doing, it would be good to add a linter for it (in another PR). flake8-alfred looks like it can only scan for exact matches, but I think git grep "def test|class Test" -- hypothesis-python/src/hypothesis is probably enough to handle this case!

One thing worth noting is that currently we have some methods which this would catch, and those aren't problems as long as they don't belong to a class named Test*. It's only top level functions that are a problem.

An alternative way we could lint this is that we could check that there are no globals in a Hypothesis test package that start with Test or test and come from a Hypothesis module and aren't the stateful testing TestCase (maybe the check here could be that they're imported with the same name as their original name - so this would fail because we call those globals TestFoo).

[Zac-HD]

git grep "^def test|^class Test" -- hypothesis-python/src/hypothesis then! If we can't do it with grep I wouldn't bother; a more complicated check will probably take more effort to maintain than it saves.

[Zalathar]

Other than that possible tweak to coverage, this all looks good to me.