I have followed the guide and tried different ways to solve this but I cant figure out how to do this properly.
https://help.prusa3d.com/article/adjusting-belt-tension-core-one-l-core-one_845048

The gantry alignment
Okay, so I have followed the steps outlined in the "Quick gantry alignment" section which seems to be another approach compared to what is described in the parent "Alignment" section: "Loosen the bolt on the corner with play less than the other." where i have loosened it completely as suggested in the quick version.

Anyway, I bent it into shape. I think... because its unclear how to test this. So far I have put the nextruder in the middle of the gantry and then gently pushed the gantry (pushing it as centered as possible) towards the tensioners looking for gaps. First I tried to make sure it was even by pushing the left side against the tensioner and a gap showed up on the other side. Doing the same thing on the right side caused a gap at the left side. Someone in the comments on the support/guide page had the same problem which then was advised by another commenter to tighten the belts somewhat. But I question that method because tightening the belts ever so slightly can possibly at this point have the undesired side effect that with the help of the belts straighten the gantry. The gantry is somewhat flimsy by design (nothing wrong here, just an observation) so I assume that some play is to be expected?

Keeping the gantry square
When it comes to the initial tightening of the belts to keep the gantry square, how should I know how much I have to tighten the bolts? There are no visible indicators. Should I just keep turning the bolts gradually and hope for the best? I've tried to use some reference guides in an attempt to make the tensioners equally loose and from that gradually tightening them as described. Problem is that the left one will become tighter sooner than the right one causing an uneven tension on the gantry. I've seen some people suggesting that you should then only continue with the right one so that the gantry is square and rely on the tuning later on.

Belt tension
Im using the "Belt tuning wizard" and I wonder how to accurately perform this test? If the top belt starts swinging at 88Hz I can keep it going up to 96Hz and at 96.5Hz it stops. Going backwards, from lets say 98Hz and down to 96Hz it will not start swinging until i reach a lower frequency, lets say 93Hz.
This made me stop the belt (by holding it still with a finger and then letting go) at each increment and let the belt start swinging by itself without the help of the previous frequency to find out the lowest and highest Hz. This narrowed the frequency range and somewhere in this range there will be a frequency that moves the belt the most...
But what to do when the belt moves about the same at lets say 92Hz as it does at 94Hz?
Am I doing something wrong here? What would the correct procedure be?

According to the guide "On CORE One L, the ideal target frequency range is a maximum of 95 Hz for the upper belt and a minimum of 85 Hz for the lower belt."
What does this mean? Anything between 85-95Hz is good enough for both belts?
Why then is the wizard so picky about some frequencies but not others? Sometimes it says "within range, nothing to do" and sometimes it wants me to adjust the tension even though both belts are within range? What am I missing?

For the core one 96Hz is optimal for the upper belt and for the lower belt it is 92Hz. There doesnt seem to be an equivalent for the Core One L.

Sorry for all the questions but I really want to understand this.

Is anyone else experiencing this problem? I would love to find our why it happened as the error message tells me to turn off the machine and contact Prusa.

The Firefox app I also in chrome and with and without API key. Is there any trick? Or does this not work? 

Sorry the screnshot is in german.

Prusa i3mk3 is giving me a thermal runaway error and it has melted the heat shroud twice already. I have replaced 2 thermistors already and I still get the same issues after doing 5 to 10 prints. Here are some pictures of what's going on and I dont know what else to do. So I have tried running the paid calibration , and it fails immediately since thermal runaway. I am using overture p p l a default settings in prusa slicer version 2.9.5. It has been frightening fine for the last 3 to 4 months.And this just started happening about a month ago.