More combination modes

[tyt2y3]

My crate currently has 10 features, it will then explode to 2^10 = 1024 combinations.
It'd be useful to be able to test some meaningful subsets of all the possible combinations.

I'd propose some modes:

one-by-one: enable each feature one-by-one
all-features: all features enabled
pseudo-random: pseudo-randomly pick N combinations
random: true-randomly pick N combinations
mix-and-match: divide features into M groups, and randomly pick R features from each group

For example, if we divide 10 features into 2 groups of 5, and picking up to 3 from each group, then:

((5 choose 3) + (5 choose 2) + (5 choose 1)) ^ 2 = 625 combinations.

After that, we can still randomly pick N from these possible combinations.

It would be a more reasonable set of combinations than picking purely randomly, as I observe that crates with many features usually divide their feature flags by function categories. As an analogy people can choose different wheels and engines when building a car.

What do you think about this?

[demmerichs]

Some thoughts on this one:

When you want to group by category/functionality, see #42. I am working on a PR that will allow to specify quite flexible rules regarding grouping and relating feature flags.

I can see the need for having different testing configurations, staying in the analogy, e.g. if we change something at the engine, we only want to test feature combinations relating to the engine. For this it would be useful, to have multiple named setting groups that change the behavior of feature set selection just by providing a group name at the CLI. This would also allow to configure a more restricted/reduced feature set for fast testing, and a complete/large feature set matrix for an important milestone testing which is allowed to run longer.

Regarding the random subsampling, I am not a big fan of that. The whole purpose of testing feature combinations and having control over it, is to give developers confidence that they did not miss a bug. Just randomly checking a few does not provide this confidence.