Thanks, Rill, but I stick to this thread, because my problems immediately vanished after my ISP initiated the monthly 5 GB with real (= UMTS) mobile broadband bandwidth some minutes ago. Before (yesterday in my timezone, before 23:00Z) I found an interesting way to maximize the failed login attempts:
1: clear cache, login, and start the Chrome "error console" (an option in "inspect this element")
2: after login the first (hanging) page loads, and the error console shows all load errors
3: if a load error affects a script (anything.js) reload the first page (goto 2), it's anyway hopeless
With this "watch error console and reload a.s.a.p." trick I arrived at about 5 reloads/minute. Out of curiosity I also tested a ping to the elgea server: Normally it's about 200ms without packet loss. At the time when the first page hangs it's 200ms up to over 3000ms with 30% packet loss. Unsurprisingly this does not work, one ore more of the critical anything.js are never loaded. For a few hours yesterday I was lucky and got two working Illy browser tabs -- all attempts to open a third Tab failed, but hey, one good tab is enough to keep the build and research queues busy.
Illy worked also for low bandwidth last week, apparently a code or server change over the weekend was a seriously bad idea for some "smallband" users.