Woke up Sunday at the correct time (barely) to get the experiment setup. Still had a bauble due to some materials being in the wrong place at the wrong time. The bauble revealed some interesting things however.
Here is the setup prior to waking up and logging in. Including the mistake. There are 4 relevant processors attached to my storage and in the storage I have the following:
1600 x Proteins
1600 x Biofuels (the mistake - lucky I had over 1600 Bacterias in a nearby spaceport)
3200 x Reactive Metals
3200 x Precious Metals
2 processors setup for Fertilizer
2 processors setup for Mechanical parts.
So I log in with 5 min to go (told you it was a little close to downtime). Rush to route the inputs. Have a problem with the biofuels not being accepted. Look at the configured schematic. Realize the Biofuels are wrong. Need Bacterias. Lucky enough bacterias are nearby. Hit submit for the existing routes which starts of the 40h of mechanical parts processing on two processors. Then I expedite the Biofuels out of the storage and the Bacterias into the storage. Route those and it submit a 2nd time. Lucky with 2 minutes to spare before downtime. So 40h reaction started at 10:56 and 20h of reactions started at 10:58...
Then I log back in at 11:35 to check what happened. Interesting the 40h reaction is proceeding as planed - currently with no losses. The 20h reaction however got insta-completed. From this I deduce that the instant completion on downtime bug is related to the most recent submit button. Whether it's relative to downtime or simply the most recent submit button is undetermined (more shit to test out - Yay!). However this is the 4th downtime where this effect has been observed when I setup the routing just before downtime. I'm beginning to suspect that it's related to the most recent submit button before downtime. Is this per planet or absolute per character? No clue (again more testing).
Again I still do not have enough of a handle to submit a bug report. Too many questions remain unanswered... Although the repeatability of the effect leads me to believe I'm getting closer to correctly understanding the mechanism.