Resiliency: Power outage is not a major data-loss risk here. Lose power during playback, you have no data at risk. Lose power during a rip, you'd probably restart the rip from scratch anyways. Individual file corruption is a relatively minor issue, since you have an off-line backup (the original discs) and lots of practice restoring from that backup. Systemic file corruption is your biggest risk, since that carries the biggest penalty (labor and downtime reripping entire collection). The best defense against that is a well-tested filesystem.
Performance: With the boot-on-demand setup, it doesn't sound like you foresee serving lots of clients. No need for anything fancy.
In your shoes, I'd go with ext4 and forget about it. I might even consider ext2.
Another thing to think about is your raid setup. If the controller fails, what's your play? How long will it take to repair or replace it? How much more value does an 18TB RAID6 have over a 12TB RAID1 stack? You are covering yourself against 2 independent drive failures, but not all failures are independent (power surge, burst pipe, rack collapse, etc) In a 3-drive failure, you lose the whole collection with RAID6, but only 1/8th of the collection (on average) under RAID1.
What you are doing is basically providing a fast cache in front of slower storage (the original discs) This isn't like running a SAN that's holding OLTP databases, VM images, or encrypted data. In your scenario, partial data loss is tolerable. That changes the design.