If there is bit-rot in a disk, which in my experience is extremely rare, and you run a correcting check parity will be incorrectly updated, note that we always recommend running non correcting checks unless sync errors are expected, like after an unclean shutdown for example, if sync errors are found without a known reason and the user has checksums for the data disks he can then first check those and then decide how to proceed.
Depends on how you do the backups, I for example use btrfs send/receive to backup all my array disks 1 to 1, the stream is checksummed all the way, if there's a checksum error on read it will abort, and as far as I understand the sent metadata also includes the checksums.
Usually you cannot have just the benefits of something, Unraid is very flexible with for example using the full capacity of different disks sizes, recovering the data from the remaining disks if you lose more disks that redundancy can save, but there are also some drawbacks, like array speed and array bit-rot fix, there are the pools where can have redundancy, but btrfs raid5/6 is not stable enough for the typical user and raid1/10 is not very efficient for large pools, soon you'll be able to create a zfs pool, of course you'll lose some of Unraid array advantages, like can't fully use different size disks with raidz and you can lose the complete pool if more disks than current redundancy can support are lost, again you cannot have everything.