chore(nix): the real vendorHash for x/crypto v0.56.0 (BUG-2851)

sha256-8L7gH7Yy5+Fig3wK2SPLYSJjcY9nF/jumQ7PATJ3RIE=

Taken from the previous commit's Nix run, which failed on the placeholder by
design:

    error: hash mismatch in fixed-output derivation
             '/nix/store/ys5igdnqzp2cdj40b27frf5f93111dvv-pad-0.15.0-go-modules.drv':
           specified: sha256-AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
              got:    sha256-8L7gH7Yy5+Fig3wK2SPLYSJjcY9nF/jumQ7PATJ3RIE=

Split from the bump rather than amended into it so this provenance is legible
in the history: the value came from a build that computed it, not from a guess
that happened to be right. There is no nix on the machine this was bumped from
— no /nix, nothing on PATH — so package.nix's own documented regeneration
command was not available locally and CI's mismatch is the only source.

The check that matters is the next Nix run on this tip: it must go green, and
a green run here is the only thing that shows the hash is correct.

nix/package.nix

@@ -61,13 +61,7 @@ buildGoModule {
 
   # Update alongside go.sum. Regenerate via:
   #   nix build .#default 2>&1 | grep -A2 'got:'
-  #
-  # PLACEHOLDER — this build is EXPECTED TO FAIL, and the failure is how the
-  # real hash is obtained. There is no nix on the box this was bumped from, so
-  # the value below comes from CI's own mismatch message on the first push of
-  # this branch. Replaced in the following commit; if you are reading this on
-  # main, something went wrong.
-  vendorHash = lib.fakeHash;
+  vendorHash = "sha256-8L7gH7Yy5+Fig3wK2SPLYSJjcY9nF/jumQ7PATJ3RIE=";
 
   subPackages = [ "cmd/pad" ];