Separation of concerns

[matatk]

As @xi suggested, it could be good to have a repo that contains just the tests (and separate repos for any related tools), then people could use just the bits they need. My intention with this project (as a whole) is to provide three things:

1. A set of landmark (and, later, heading and article) test cases.
2. Convenience code such as a runner that projects could use out-of-the-box.
3. Benchmarking tools to compare scanners (including assistive technologies).

I'd like to make it very easy to produce the benchmark results, e.g. they should come "for free" if the convenience test runner is used, but I agree there's no need to host all of that code in this repo. Further, there is quite a bit of variation in how different tools (non-JavaScript ones in particular) might be tested, so having all those dependencies in here is also not very UNIX-like (nor npm-like :-)).

Therefore I propose to refocus this repository so that it contains only:

• The test cases (separate and combined).

• README content relating to the standards support.

• A minimal "index.js" that consumers can use to quickly access the fixtures and expectations.

  I'm thinking of a data structure that provides the expectations and pointers to the HTML file(s). An alternative would be to provide the contents of the HTML file(s) as strings.

  Of course, consumers could just ignore the above code and work out the file path into the "node_modules" directory for themselves, though if it was done minimally and usefully, some assistance with iterating over the tests and expectations could be nice.

Another question: is "fragment" a better word than "combined" to refer to the combined HTML fixtures? Will think on that.

There would be another repo for the test runner and continuation of the development of the benchmarking code.