Allow empty argument in functional pseudos

[bramus]

Fixes #219.

Marking this PR as draft for now, as I’m not sure what to do nth-child and its variants. Should relaxed parsing – allowing an empty argument – also be allowed there? If so, it’s a matter of changing the fixture to not mark "" as invalid.

[lahmatiy]

@bramus

If so, it’s a matter of changing the fixture to not mark "" as invalid.

A failed test is a little weird. It tests parsing <an-plus-b> but through :nth-child() pseudo. So empty string is invalid for <an-plus-b>, but that's fine for :nth-child() content. I think you may change the test to ignore empty string case in fixture.invalid.forEach block.

Everything else looks good to me 👍

[coveralls]

Coverage increased (+0.0004%) to 98.932% when pulling 96d9ded on bramus:allow-empty-argument-in-functional-pseudos into 593bf37 on csstree:master.

[bramus]

Updated tests to reflect resolution as per #220 (comment)

[lahmatiy]

@bramus Thank you!