Current build timeout configuration leads to unnecessarily aborted builds
In e47abfe7, we added a build timeout to the CI process. The way this timeout is configured also takes queuing time into account. This means if multiple jobs are queued, most of them will eventually fail, as as successful build will currently take 2h on its own.
I'd like to move the timeout into the individual build stages (where we can reasonably estimate a "normal" build time). Maybe this approach could be combined with an overall timeout.