zfs: 0.8.2 -> 0.8.3

[NeQuissimus]

Motivation for this change

Just trying the patchset that may become zfs 0.8.3 (see openzfs/zfs#9776)

Since there was a minor change in our derivation, I figured I'd send the change up.
For the moment, do not merge, this is beyond unstable :D

I am running this right now with kernel 4.19.95-hardened and things seem smooth.

Things done

• Tested using sandboxing (nix.useSandbox on NixOS, or option sandbox in nix.conf on non-NixOS linux)
• Built on platform(s)
  • NixOS
  • macOS
  • other Linux distributions
• Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
• Tested compilation of all pkgs that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
• Tested execution of all binary files (usually in ./result/bin/)
• Determined the impact on package closure size (by running nix path-info -S before and after)
• Ensured that relevant documentation is up to date
• Fits CONTRIBUTING.md.

[NeQuissimus]

I believe this needs more work for 5.x kernels. I have not had a chance to try that.

[NeQuissimus]

Updated the changes to work with kernel 5.4.11-hardened
openzfs/zfs#9089 has a change that bends a systemd unit to /dev/null/ but something else in the build tried to run sed -i against that, which fails.
It sounds like the change is Debian-specific, so I am partially reverting that particular diff.

Edit: Confirmed working with 4.19.95-hardened as well. Now my poor m3 laptop CPU is done building kernels :D

[NeQuissimus]

The biggest thing here is that the ZFS-native encryption now uses hardware instructions. My SSD is finally reading faster than 20 MB/s again :D

[NeQuissimus]

Updated to use the actual 0.8.3 tag that was just released.
I built my NixOS config with this and here I am, so things seem to work :D

[wizeman]

Looks good to me, but can you take a look at #78415?
We might want these two PRs to be merged together into one, as each one of them seems to have fixes the other doesn't.

[NeQuissimus]

I think that PR may actually be a cleaner way of resolving the issues I've seen. Let's move the discussion there.