[meta] Tests for abpcrawler

[sergz]

Background

Although the crawler by itself is some kind of tests and tests of the tests look not so good, the reasons of recent changes are proof that we need some kind of tests. Especially as soon as we start to run it automatically and add gathering of more attributes we will definitely need tests.

I would like to even include running of them into each run of the crawler. So far, I'm mainly manually looking at the file names of the output and at the size of those files. In addition sometimes I'm looking into the files, to control that all information (filters, headers and so on) is actually there. Technically, we can already test almost everything what it is doing right now. We can have a couple of our artificial test pages (served by current server in run.py) and check that the output is correct.

• Serialized output, headers, filters, AA criteria (e.g. size of interesting elements) - compare with expected value.
• Screenshots - generally speaking the actual rendering can vary (see also #3941), so we probably won't be able to simply compare the result with the expected value, but we can at least ensure that the size has a reasonable value and the screenshot is not a single color image (e.g. black).

What to change

Let's firstly discuss what and how should be done.

[trev]

Mass-closing all bugs in the Extensions for Adblock Plus module, these extensions no longer work as of Adblock Plus 3.0 / Firefox 57.