Unfortunately, it appears our "fix" was that we had mapped command parameter IDs incorrectly (e.g. if a parameter was a position/string/integer). After these were mapped correctly, the issues reappeared. This does point to the issue lying somewhere with an ID, though.
I can confirm that command loading is SIGNIFICANTLY better in 1.16.100. Under the same circumstances as in that video, there was maybe a half-second pause. With more commands it can take three-to-five seconds, but it's clear that the situation is far better. I'll leave it up to someone else to determine if this is still a significant issue.
Same error occurs on 1.16.2 client as well.
Thanks for reopening this. This is just a report that the issue still occurs on rc2.
Hello! There's no evidence to prove this has been fixed in 1.17. I hope to provide more solid replication steps in the future.