Fix/join datasets

[tomwayson]

this is a WIP, everything is working, just need to iron out a few wrinkles and squash

resolves #406

[tomwayson]

This is why Jest is dumb

[benstoltz]

So visually everything looks okay with a multi-series dataset, however if there was a missing value it doesn't fill in with a 0, it instead fills in with a blank value. The result looks something like this:

Now if it was a record of 0 then it would look like (bear in mind the middle value has no val assigned to it):

I suppose the question is, do we want this type of behavior, or do we want it filling in a record of 0?

[tomwayson]

Glad you brought that up. I meant to include screenshots and explain the same. I actually had the code to add those, but the only way I could think to do that requires another loop through each value field for each dataset for each row, so I thought the above behavior was sufficient, and it wasn't worth the overhead

This morning I had the thought that we could add an option like fillInMissing, but where, passed to show()? On the chart definition itself, like under dataset? Either way, that's something I think we can add later as a feature, but I do think it's worth taking another look at the API I've made and in this PR and thinking about how we would pass that and/or other options. For example, we could have an option to force calling getAttributes() on every row instead of just checking the first one for the extremely rare situation where you have mixed features and non-features.