Caching, fetching and inspection

[tornqvist]

This is quite the rewrite and I'm fully aware it may not be well received. I was about to create something along the lines of nanographql before I knew it even existed. But as I started looking into parsing GraphQL queries with RegExp I realized we can do much more than just deriving the operation name. This did however bring a significant increase in size, from 15 LOC to ~200, so it may not be legitimately "nano" anymore.

Since I'm not currently a user of nanographql I'd really appreciate some feedback from people using it. If the changes are not welcome I'll just publish it under a different name. Alternately, the cache and fetching bits could be split out to another package, e.g. choo-graphql or something. However, these changes do address all issues and PRs currently open in this project.

These changes address:

• Expressions: strings, and even other queries (fragments), can be interpolated to create dynamic queries (Expressions support #1)
• Inspection: the return value from a query is an object with methods to serialize it in many different ways (toString, toJSON) and even iterate over its properties (variables must be a object #7)
• Fragments: fragments are supported and can even be interpolated from other queries (support fragment? #8)
• Caching: results are cached per query, which allows for maintaining state between queries and mutations (What about some cache mechanism? #11)
• Update standard: standard recently introduced the no-var rule in a major update

See README for complete documentation of new functionality.