style: Make Invalidation work in terms of a dependency, not a selector.

That way we can look at the parent dependency as described in the previous
patch. An alternative would be to add a:

    parent_dependency: Option<&'a Dependency>

on construction to `Invalidation`, but this way seems slightly better to avoid
growing the struct. It's not even one more indirection because the selector is
contained directly in the Dependency struct.

Differential Revision: https://phabricator.services.mozilla.com/D71422

components/style/invalidation/element/document_state.rs

@@ -6,6 +6,7 @@
 
 use crate::dom::TElement;
 use crate::element_state::DocumentState;
+use crate::invalidation::element::invalidation_map::Dependency;
 use crate::invalidation::element::invalidator::{DescendantInvalidationLists, InvalidationVector};
 use crate::invalidation::element::invalidator::{Invalidation, InvalidationProcessor};
 use crate::invalidation::element::state_and_attributes;
@@ -65,6 +66,11 @@ where
     E: TElement,
     I: Iterator<Item = &'a CascadeData>,
 {
+    fn check_outer_dependency(&mut self, _: &Dependency, _: E) -> bool {
+        debug_assert!(false, "how, we should only have parent-less dependencies here!");
+        true
+    }
+
     fn collect_invalidations(
         &mut self,
         _element: E,
@@ -81,10 +87,14 @@ where
 
                 // We pass `None` as a scope, as document state selectors aren't
                 // affected by the current scope.
+                //
+                // FIXME(emilio): We should really pass the relevant host for
+                // self.rules, so that we invalidate correctly if the selector
+                // happens to have something like :host(:-moz-window-inactive)
+                // for example.
                 self_invalidations.push(Invalidation::new(
-                    &dependency.selector,
+                    &dependency.dependency,
                     /* scope = */ None,
-                    0,
                 ));
             }
         }