I personally haven’t been able to replicate this at all, and it hasn’t shown up much since it was happening to you and a couple other users.
In terms of a workaround, you could use the AMP API to attempt to start things, but honestly cron would be the easier choice in this scenario.
I do wonder however, if this happens to new instances on your machine, or if it’s just select instances created when PreStart was being buggy or something strange like that.