red

It's the same for both local and cloud drives being removed from a pool: "Normally when you remove a drive from the pool the removal process will duplicate protected files before completion. But this can be time consuming so you can instruct it to duplicate your files later in the background."
So normally: for each file on the drive being removed it ensures the duplication level is maintained on the remaining drives by making copies as necessary and only then deleting the file from the drive being removed. E.g. if you've got 2x duplication normally, any file that was on the removed drive will still have 2x duplication on your remanining drives (assuming you have at least 2 remaining drives).
With duplicate files later: for each file on the drive being removed it only makes a copy on the remaining drives if none already exist, then deletes the file from the drive being removed. DrivePool will later perform a background duplication pass after removal completes. E.g. if you've got 2x duplication normally, any file that was on the removed drive will only be on one of your remaining drives until the background pass happens later.
In short, DFL means "if at least one copy exists on the remaining drives, don't spend any time making more before deleting the file from the drive being removed."
Note #1: DFL will have no effect if files are not duplicated in the first place.
Note #2: if you don't have enough time to Remove a drive from your pool normally (even with "duplicate files later" ticked), it is possible to manually 'split' the drive off from your pool (by stopping the DrivePool service, renaming the hidden poolpart folder in the drive to be removed - e.g. from poolpart.identifier to xpoolpart.identifier - then restarting the DrivePool service) so that you should then be able to set a cloud drive read-only. This will have the side-effect of making your pool read-only as well, as the cloud drive becomes "missing" from the pool, but you could then manually copy the remaining files in the cloud poolpart into a remaining connected poolpart and then - once you're sure you've gotten everything - fix the pool by forcing removal of the missing drive. Ugly but doable if you're careful.

Not currently, but I definitely do keep on bringing it up. 

CloudDrive itself operates at a block level, so it isn't aware of what files in the file system are being written to or read by the applications on your computer. Much like the firmware for a HDD or SDD is unaware of that information as well. So, that is to say, that it isn't possible via CloudDrive. Windows Resource Monitor or Process Explorer or another tool to look at Windows' file system usage would be required--as it sounds like you discovered.

Sorry about the tinted screengrab (f.lux). What's this about?

Hello,
So I'm running a 100TB plex server with only 60TB of local storage and would like to duplicate only certain folders from my clouddrives to local. So for example let's say I want 1000 of my 3000 movies on the local storage. As far as I can tell (and what I've been doing) is going one by one through each folder and setting duplication. Even just being able to select multiple folders or having a config file would be a great improvement to the current system.

For anyone reading this, Plex is smart enough if you move an existing Library file from Path A to Path B to realize you actually moved it, and that it's not a new duplicate. So while what @josefilion did worked, it really doesn't need to be done that complexly, or at one go. I recently moved from one huge folder with subfolders to one with A-C, B-D etc subfolders because it was taking Windows a long time to list the contents and I basicly just grabbed a few hundred folders at once with cut & paste to the new subfolders and me having "Detect changes" enabled in Plex made it pick up the moved files almost instantly.
So basicly: No need to turn off Plex. Just move your files how you want as long as they are not currently being played / scanned and you move them within the Library root folder.