Add "regaining document full activity steps"

These steps will be run when a document regains full activity after restoring it from BFCache. This does not currently introduce any behavior changes within the HTML spec, but allows other specs to define the steps they will take when a document becomes fully active again, e.g. send updates to the renderer about the current Geolocation state, etc.

Fixes whatwg#8872.

source

@@ -94267,6 +94267,39 @@ location.href = '#foo';</code></pre>
   href="#note-bfcache">bfcache</a>, i.e., after it has been made <span>fully active</span>
   again.</p>
 
+  <ol>
+   <li>
+    <p>Run any <span>regaining document full activity steps</span> for <var>document</var> that are
+    defined by this specification and <span>other applicable specifications</span>.</p>
+   </li>
+
+   <li>
+    <p>If <var>document</var>'s <span>current document readiness</span> is "<code
+    data-x="">complete</code>", and <var>document</var>'s <span>page showing</span> flag is false,
+    then:</p>
+
+    <ol>
+     <li><p>Set <var>document</var>'s <span>page showing</span> flag to true.</p></li>
+
+     <li><p><span>Update the visibility state</span> of <var>document</var> to "<code
+     data-x="">visible</code>".</p></li>
+
+     <li><p><span>Fire a page transition event</span> named <code
+     data-x="event-pageshow">pageshow</code> at <var>document</var>'s <span>relevant global
+     object</span> with true.</p></li>
+    </ol>
+
+    <!-- an interesting thing to test would be to traverse back during onload, before the first
+    pageshow has fired, and then to traverse forward again, and see if we get _two_ pageshows. If
+    so, it indicates that browsers don't have a "page showing" flag like this and that the history
+    traversal task source has a higher priority than the DOM manipulation task source. -->
+   </li>
+  </ol>
+
+  <p>This specification defines the following <dfn export>regaining document full activity steps</dfn>.
+  Other specifications can define more. Note that these steps must not run script.<br>
+  Given a <code>Document</code> <var>document</var>:</p>
+
   <ol>
    <li id="history-autocomplete"><p><span data-x="list iterate">For each</span>
    <var>formControl</var> of form controls in <var>document</var> with an <span>autofill field
@@ -94293,28 +94326,6 @@ location.href = '#foo';</code></pre>
      <var>suspendDuration</var>.</p></li>
     </ol>
    </li>
-
-   <li>
-    <p>If <var>document</var>'s <span>current document readiness</span> is "<code
-    data-x="">complete</code>", and <var>document</var>'s <span>page showing</span> flag is false,
-    then:</p>
-
-    <ol>
-     <li><p>Set <var>document</var>'s <span>page showing</span> flag to true.</p></li>
-
-     <li><p><span>Update the visibility state</span> of <var>document</var> to "<code
-     data-x="">visible</code>".</p></li>
-
-     <li><p><span>Fire a page transition event</span> named <code
-     data-x="event-pageshow">pageshow</code> at <var>document</var>'s <span>relevant global
-     object</span> with true.</p></li>
-    </ol>
-
-    <!-- an interesting thing to test would be to traverse back during onload, before the first
-    pageshow has fired, and then to traverse forward again, and see if we get _two_ pageshows. If
-    so, it indicates that browsers don't have a "page showing" flag like this and that the history
-    traversal task source has a higher priority than the DOM manipulation task source. -->
-   </li>
   </ol>
 
   <p>To <dfn>try to scroll to the fragment</dfn> for a <code>Document</code> <var>document</var>,