Init bakke (!87)

New backup server just dropped!
This server is awfully slow, and the mdraid setup is awfully slow, and I doubt that this will be a good experience, but we now have a backup server again?

- Tried Disko and nixos-anywhere
- Tried using mdraid
- Found that md is ancient and bad
- Found that disko is 100% extra steps, and a lot more complicated and noisy than just formatting your disks yourself
- Found that systemd-boot doesn't support mdraid
- Found that we probably don't need to mirror the boot partition :)
- Found that old hardware is slow
- Found that old hardware can have poor support for iPXE with UEFI, and might do weird BIOS stuff on you when you least expect it
- Reaffirmed that zfs is love

Current disk layout:
- mdraid for boot/root disk
    - 4TB WD Red with 500MiB ESP with systemd-boot, Remaining mdraid - Old?
    - 4TB WD Red with 500MiB Unused partition, Remaining mdraid - Old?
- zfs pool "tank" for the actual backup data
    - 8TB Toshiba MG08 - New
    - 8TB Exos 7E10 - New

TODO:

- Document the death of Toriel on the wiki
- Document Bakke on the wiki
  - ... describing the poco loco disk layout
- Start backing stuff up
  - Restic? Borg? Rsync?
  - Make backup retention policy and zfs snapshot system
  - Document backup procedures

Reviewed-on: https://git.pvv.ntnu.no/Drift/pvv-nixos-config/pulls/87
Co-authored-by: Felix Albrigtsen <felix@albrigtsen.it>
Co-committed-by: Felix Albrigtsen <felix@albrigtsen.it>

flake.nix

@@ -104,6 +104,11 @@
       stableNixosConfig = name: extraArgs:
           nixosConfig nixpkgs name ./hosts/${name}/configuration.nix extraArgs;
     in {
+      bakke = stableNixosConfig "bakke" {
+        modules = [
+          disko.nixosModules.disko
+        ];
+      };
       bicep = stableNixosConfig "bicep" {
         modules = [
           inputs.matrix-next.nixosModules.default