I'd guess the SSD's are connected via USB and likely the same internal hub/port complex, and the PSU (or the hub/port complex) can handle one drive being connected, but not two, resulting in disconnects when power draw is too high.
None of that looks like a debug log, so thanks and good luck.
Update the script to use /dev/sda and see if that resolves the issue? - the hdparm command targets a disk device and /dev/sda2 is a partition on a disk, not the disk.
Has this been added to nightly builds?
No, it's only present in the RPi5 image in my test share (as linked above).
Correct, it has nothing to do with HDR playback and reducing brightness on the OSD layer.
umount -f /dev/nvme1n1
parted -s /dev/nvme1n1 mklabel gpt
parted -s /dev/nvme1n1 mkpart primary ext4 0% 100%
mkfs.ext4 -L DISKLABEL /dev/nvme1n1p1Something like that ^ should do it - I'm typing commands from memory so syntax might not be 100% correct. The commands force unmount any existing partition (might fail if there isn't one) and then create a GPT partition scheme, create an EXT4 partition that's sized to be 100% of the disk, and then format the filesystem with "DISKLABEL" as the disk label. On reboot the drive should be auto-mounted under /var/media/DISKLABEL.
I have zero interest in what works/doesn't with CE because the entire display pipeline is different and comparisons are irrelevant. If you want to troubleshoot LE, post the debug log requested else nobody can help.
No further support due to presence of banned add-ons: https://kodi.wiki/view/Official:Forum_rules/Banned_add-ons
No idea what the issue is but you can install evtest via the "System Tools" add-on in the LE repo.
chewitt is this something that will help here? once things pick up and its added?
IIRC the DevCon discussion was about work done by OSMC to implement broader support for DV using 'open' capabiltiies and thus avoid the need for closed-source dovi blobs and DV license fees. That work proves it can be done but isn't easily transferred to other hardware platforms as it's all Amlogic vendor kernel/uboot specific.
CEC is working on v12.0.1 but stops working on v12.0.2 for whatever reason
The kernel used in LE 12.0.0 and 12.0.1 supports CEC on both HDMI ports but there are some issues when multiple CEC-capable HDMI devices are active at the same time so RPi devs dropped that capability. The kernel in LE 12.0.2 reverts to the behaviour of LE10/LE11 where CEC is only supported on the HDMI port nearest the USB-C power connector.
Hard to comment without a Kodi debug log to show what's going on 
If the USB drives have EXT4 partitions you can use "e2label" to change disk labels. If they are NTFS rename on Windows.
Kodi has a simple "shutdown timer" which you can set via the power icon in the top-left corner of the home screen. Once the timer has been set the system shuts down (without prompting to remain awake) when the timer value is reached. So if you are confident of falling asleep in ~20 mins, set the value to 30 or 40 mins and the system will be off shortly after you are.
If you are playing local video media or some form of PVR or internet stream, the system remains in an active state for the duration of the media or stream and the screensaver will not show until x minutes after the system is seen to be inactive.
Detecting when you fall asleep and sending a remote shutdown command X mins later would be a fun RPi + sensors project 
NB: Most TV's have a built-in sleep timer function. I have elderly relatives that have their TV configured for 60 mins. It prompts to reset/continue by hitting OK on the remote and if not pressed (because they dozed off) it shuts down.
I've been using LE13 images for 12+ months without any real-world issues, albeit on an RPi5 and not something that uses Generic with an AMD card. It would be more useful to triage on the current development release since that will receive updates. LE12 is less likely to see another release.
The first thing I'd do is update to an LE13 nightly to see if the newer kernel resolves the origina (radeon) issue?
Latest is Nexus not Omega
If the main repo hasn't reveived new commits but we've been chasing a CI build error/issue resulting in multiple CI runs it can happen.
