Support :host and :host-context pseudo class selectors

[bramus]

These are defined in https://w3c.github.io/csswg-drafts/css-scoping/#host-selector

• :host
• :host(<compound-selector>)
• :host-context(<compound-selector>)

With this PR, CSSTree will now parse them, instead of returning a Raw AST.

Limitations of this PR:

• CSSTree accepts a <complex-selector> as the argument to :host() and :host-context(). I don’t consider this blocking to this PR as CSSTree has the same problem with ::slotted() for example. See issue CSSTree does not differentiate between <compound-selector> and <complex-selector> #214.
• :host-context should be invalid yet CSSTree does not consider it as such. I don’t consider this blocking to this PR as CSSTree has the same problem with :has vs :has(). See issue Pseudos that require an argument parse fine, although they shouldn’t #215.

[coveralls]

Coverage increased (+0.0002%) to 98.931% when pulling 449cc41 on bramus:host-pseudos into debb7c5 on csstree:master.

[lahmatiy]

@bramus Thank you for your effort!
I commented #214 and #215, that's not a bug but a feature.
It would be nice to add a test for :host. I believe it can be parsed. However, let's be double sure it works. Everything else looks good to me.

[bramus]

Done.

[lahmatiy]

Thank you, @bramus!