Migrate all ParserTest fixtures to the gherkin testdata format

[stof]

Now that we have support for using custom tests written using the cucumber gherkin testdata format, all our existing fixtures used by ParserTest should be converted to the gherkin testdata format.

This will unblock #362 (and might unblock contributing some cases upstream to increase coverage)

[acoulton]

Thanks @stof - I had the same thought just now as well and was just about to post it 😆

I'd also like to generate gherkin-testdata-format expectations for the upstream examples we know we parse differently (rather than just skipping them in legacy mode) which might allow us to delete some of our own examples already and will give better confidence that we've not changed any legacy mode parsing. (this bit done slightly differently in #392)

[Amoifr]

@acoulton it has been a while since this and #362, is the fixture migration still on your list? If not, I would be happy to take the ParserTest fixtures conversion to the gherkin testdata format (I have been around this area recently with Behat/Behat#1863). No rush either way, just checking before touching anything.

[acoulton]

@Amoifr it's still on my list, but my list is loooong and my time is short 😆

If you would like to take a look at this I would be delighted! Thanks :)

[Amoifr]

Thanks @acoulton! PR is up: #427. All 34 fixtures made it to extra_testdata/good with expectations generated by the official gherkin-javascript parser, and only 7 diverge, all in legacy mode only (multiline titles and description trailing whitespace), now documented with recorded variants. ParserTest no longer touches YamlFileLoader, so #362 is unblocked. Your #322 was a great map to follow!