The first time a `slotchange` event is registered, we setup a SlotChangeMonitor
on the page. This uses a global (ugh) MutationEvent to detect slot changes.
We could improve the perfomance of this by installing a MutationEvent per
custom element, but a global is obviously a lot easier.
Our MutationEvent currently fired _during_ the changes. This is problematic
(in general, but specifically for slotchange). You can image something like:
```
slot.addEventListener('slotchange', () => {
// do something with slot.assignedNodes()
});
```
But, if we dispatch the `slotchange` during the MutationEvent, assignedNodes
will return old nodes. So, our SlotChangeMonitor uses the page scheduler to
schedule dispatches on the next tick.
Refactors some of the module loading logic. Both normal modules import and
dynamic module import now share more of the same code - they both go through
the slightly modified `module` function.
Dynamic modules now check the cache first, before loading, and when cached,
resolve the correct promise. This can now happen regardless of the module
loading state.
Also tried to replace some page arenas with call arenas and added some basic
tests for both normal and dynamic module loading.
chromedp expects the nodeId starts to 1.
A start to 0 make it enter in infinite loop b/c it expects the Go's
default int, ie 0, to be nil from a map to stop the loop.
If the 0 index is set, it will loop...
exit_when_done is pretty much a sneaky way to get CDP knowledge into the page.
exit_when_done == true means "this isn't a CDP session".
extra_socket is another sneaky weay to get CDP knowledge into the page. When
we get an `extra_socket` message it means "Return control to the CDP server".
Therefore it should be impossible to get an `extra_socket` message (return to
CDP) when `exit_when_done == true` (this isn't a CDP session).
Karl Seguin