Sync DatoCMS w/ Algolia

[dannymidnight]

This is a slight variation on #52 by pulling the data from DatoCMS prior to indexing. Why do this? It's a little roundabout, but it's a quick way of inferring the parent -> child facets for testing out querying.

Now a search document contains the following:

{
  name: "CI for Ruby",
  languges: ["Ruby"],
  use-cases: ["CI"],
  ...
}

as opposed to:

{
  name: "CI for Ruby",
  categories: ["Ruby", "CI"],
  ...
}

Some alternatives to this approach:

• Structure the data like this in frontmatter and bail on categories
• Reference the categories from within this repository https://github.com/buildkite/templates/pull/51