I’d guess this was a consequence of a strategy decision that went against the plan (probably an expectation mismatch with users). I think the large part of the problem is the public GitLab instance is (was?) the beta environment, so a lot of issues popped up constantly while people were evaluating the options. In our group we are all volunteers, so when downtime impacts the little time we can dedicate (push code, review, deploy) it’s a fairly big deal.
Our python group (pyslackers.com) runs in the open as much as we can, so for us we went where the exposure and ease to be involved is (Github), even though from a raw feature perspective we liked what we had on GitLab when it worked.