"aborted flag" for non-network errors

[annevk]

It looks like we sometimes set the "aborted flag" on responses that are not network errors. I guess that should be possible to account for late aborted fetches. We might want to point this out more explicitly?

https://fetch.spec.whatwg.org/#dom-global-fetch only branches on "aborted flag" and not "aborted network error". Since setting the aborted flag is not queued as a task it seems that it's possible for a race condition there to cause something to throw while there was actually a response. At the very least I think we should branch on "aborted network error" in "To process response for response".

cc @jakearchibald

[annevk]

@nidhijaju perhaps this is something you can look at as well? It might not be too hard to fix at this point.