It is hard to say anything definitively based on runDisney seemingly always changing how they do things. Over the years, there have been examples of ways to do something only to see the counter done the following year. With that being said, this is my best guess as to how to handle the situation.
View attachment 331109
If runDisney continues to use the smaller number of corrals with larger number of people in them, then it likely won't matter whether you do or don't submit a POT when your POT is just slightly slower than the cutoff requiring it. You can see that corral E in 2018 was a "Time Required" of 2:31-2:45. In order to have a time in that range, it was required to submit a POT. My estimation is that it required a 10k POT of about a 1:14. With that being said, it's merely an estimation on my part using a Riegel R of 1.06 which is standard. We'd all love for runDisney to just share what calculation they use or what their race equivalent cutoffs for the 10k or 10 mile distance for HM registration is. Now if you were to instead submit a POT just slightly slower than a 2:45 race equivalent (say a 1:20 10k which is a 2:56-2:59 HM based on a Riegel of 1.06 to 1.08), then you would likely be in corral F (in 2018). This would be because your POT says 2:46-3:00 based on this conversion. If you were to submit no POT and just an estimated finish time of 2:46-3:00, then you'd also likely in up in corral F (in 2018).
Now, the important part. The estimated finish time is a critical registration spot. If you were to have a 1:20 10k POT (and thus a 2:56-2:59 HM equivalent), and yet you stated your estimated finish time was a 3:30 HM, then you would be placed in corral H. runDisney will use your self-seeded estimated finish time as a trump to the POT if you submit it that way. You're essentially telling runDisney while I am faster (per my POT) I plan to take the race slower so please seed me with the slower than my POT people. As an extreme example, if I had a POT that was a 1:30 HM, and yet on my registration said my estimated finish time is 3:30 HM, then I would likely be placed in corral H and not in corral A. This does not work in reverse (you can't have a 2:30 HM POT and estimated finish time of 1:30 HM to get into corral A).
Now where things got messy was when runDisney used to use a larger number of corrals like in 2017.
View attachment 331112
Here you can see a large subsection of "=2:50" corrals. From the examples I've read online, if people had a POT better than the cutoff (which I believe was 2:50 for that year?) then they were in corral H or better. If they had a POT just slightly slower than 2:50 (say a 2:51 race equivalent HM), then they ended up in corral L or worse. The people who ended up in corrals I, J, and K were the people who entered an "estimated finish time" of 2:50 but provided no POT. The only information available to runDisney for seeding those people was the estimated finish at 2:50 and thus they were seeded ahead of those with a 2:51 POT.
Now we can't say definitively how runDisney will handle the 2019 registrations. But in both 2017 and 2018, a person whose POT was just slightly slower than the cutoff was better off (or equal) not submitting a POT and just putting an estimated finish time right behind the cutoff (like 2:46-3:00).
I know that was a lot, so please let me know if something doesn't make sense.