Support data synthesis strategies for joint distributions

[cosmicBboy]

Is your feature request related to a problem? Please describe.

Currently, data synthesis strategies are apply to single columns or indexes and doesn' support dependencies across two or more columns. It would be nice to be able to support such strategies, however this would introduce considerable complexity to the Check and strategy interface.

Need to come up with a reasonably simple way of specifying strategies that operate on multiple columns. I anticipate needing to implement first-class Check objects that explicitly know about multiple columns, see #75 and #165

[Zac-HD]

This is going to be totally intractable for more than the simplest cases, and have some complicated interactions with e.g. #404 even for those. I'd recommend pushing the problem back onto users and recommending e.g. .map() or @st.composite, at least for now 😕

[cosmicBboy]

cool, I'm thinking of providing some built-in strategies for those simple cases and then providing an interface for custom strategies for users.