nixos/tests/pleroma: fix DB provision race condition

We originally used a systemd preStart command to provision the DB. We
recently extracted this to a proper systemd unit (to go around the
timeout issue). This created a race condition in the VM test.

Restarting explicitely the db provision unit to make sure the
migrations are applied.

nixos/tests/pleroma.nix

@@ -271,6 +271,7 @@ import ./make-test-python.nix (
         pleroma.succeed("provision-db")
         pleroma.wait_for_file("/var/lib/pleroma")
         pleroma.succeed("provision-secrets")
+        pleroma.systemctl("restart pleroma-migrations.service")
         pleroma.systemctl("restart pleroma.service")
         pleroma.wait_for_unit("pleroma.service")
         pleroma.succeed("provision-user")