The root of your problem:
Mar 2 19:23:54 NAS emhttpd: shcmd (88): mount -t btrfs -o noatime,space_cache=v2,discard=async -U 5e1bc338-27cf-4a34-8163-3495a924be67 /mnt/cache
Mar 2 19:23:54 NAS kernel: BTRFS info (device sdf1): turning on async discard
Mar 2 19:23:54 NAS kernel: BTRFS info (device sdf1): using free space tree
Mar 2 19:23:54 NAS kernel: BTRFS info (device sdf1): has skinny extents
Mar 2 19:23:54 NAS kernel: BTRFS info (device sdf1): bdev /dev/sdf1 errs: wr 0, rd 0, flush 0, corrupt 203, gen 0
Mar 2 19:23:54 NAS kernel: BTRFS info (device sdf1): bdev /dev/sdq1 errs: wr 0, rd 0, flush 0, corrupt 203, gen 0
Corrupt on a BTRFS pool is usually because of memory issues. Run Memtest for a minimum of a single pass (ideally for a couple of passes)
This has trickled down to the docker.img being effectively read-only