Gatherers should have ids

Currently we use filenames to refer to gatherers within the project. And we also publicly display the their classes `[[ClassName]]` like in this message:

> status Loading page & waiting for onload +332ms HTTPRedirect, HTMLWithoutJavaScript

IMO there's benefit to each gatherer providing an id and a readable name. (As Audits do with with `.name` and `.description` respectively)

Possible uses:

* use instead of [[classname]] when logging our loading/gather activity to the user
* For the [--do=GAR proposal](https://github.com/GoogleChrome/lighthouse/issues/1806#issuecomment-297843024), know what to name the `/artifacts/` files on disk

Adding this mostly means adding a `static get meta() {` on the gatherer classes, returning an object with some properties (a la audits). Also a test in gather-runner to assert these exist.

We can put them to real use in a 2nd PR.