Hey mate, the way I handled it was just to update the image to someting with a gauge scale, an icon (little CPU for instance) and a name, and then leave the value for the actual needle, it takes around 2-3 seconds to update the screen anyhow.
For easy to notice alerts, I just update the background LED to a gradient, eg, my base BG is white, and then if it hits a threshold of say 50, it goes yellow, then between 50 and 80 it fades between, yellow, Orange and red. That has been working well for me
I have not had a chance to check out the beta yet (I am working with their support to resolve this timeout issue I am having), sorry, but I see @osxdude has, and it looks amazing, just a note, the way the updates seem to work, is the easing sets the max they can move, and how often etc, default is every 50ms but max 2% I think, so if your CPU usages goes from 10-50, and you update every 500ms, it will update max 2% every 50ms until your next input... but you can adjust the easing settings, both the ms between updates, the % amount etc, which can be set in the web UI (kinda) or using the API, if that helps, you can set the max % to 100 (I have) but it gets pretty erratic
Edit: Just to add on some of your points.
2, I have 6 gauges (Sorry :D), but for testing I am happy to try 4, I think I will be getting my hub RMA'd soon so I will let you know.
5, I also had issues getting in contact with him, you might find better support from streacom support directly, they have been VERY responsive, and helpful, they seem to understand the product, but they do not seem to have direct control over the software unfortunately.
6, The way their demo and how I have handled it, I just set the needle to 0 and LED to off when closing the app, because otherwise if you shutdown they stay on in some cases, but leaving the lcd as a static image seems perfectly acceptable (Preferred in my case).
And just a note, I am not 100% sure on the quality, model etc of their display, but most eink displays can handle 50-100 million updates before they start to fail, and they will more than likely fail in other ways before that, so I wouldn't be too concerned about accidentally updating too frequently etc.