Expand TSML documentation

[lindholc]

The most consistent feedback we got from AGU was about documentation regarding adding datasets, which in the absence of a tool means writing TSML.

We already have a simple tutorial for time series data (#55). One idea expressed was to have similar tutorials for different patterns and shapes of data, like serving data stored as a collection of granules or spectral data.

[sapols]

Awesome. This will go a long way toward making LaTiS more usable.

[dlindhol]

I'm hoping we will be using LaTiS 3 when we release this. Not sure how much effort to put into TSML vs FDML. Fortunately there is a lot of overlap. We might want to start by being explicit in the model definition instead of relying on TSML's implicit Functions for "time" and "index" domains.

[dlindhol]

With v3 we can also use other forms of metadata. It struck me that we could statically define the info response and use it to provide the metadata instead of FDML. The other HAPI developers might feel better about enabling data providers to use that form.