llama-cpp-{cuda,rocm,vulkan}: add nixpkgs-update: no auto update

Just `llama-cpp` itself should be updated instead.

Prevents PRs like:
https://github.com/NixOS/nixpkgs/pull/560287
https://github.com/NixOS/nixpkgs/pull/560286

Signed-off-by: Ethan Carter Edwards <ethan@ethancedwards.com>

pkgs/by-name/ll/llama-cpp-cuda/package.nix

@@ -1,3 +1,4 @@
 { llama-cpp }:
 
+# nixpkgs-update: no auto update
 llama-cpp.override { cudaSupport = true; }

pkgs/by-name/ll/llama-cpp-rocm/package.nix

@@ -1,3 +1,4 @@
 { llama-cpp }:
 
+# nixpkgs-update: no auto update
 llama-cpp.override { rocmSupport = true; }

pkgs/by-name/ll/llama-cpp-vulkan/package.nix

@@ -1,3 +1,4 @@
 { llama-cpp }:
 
+# nixpkgs-update: no auto update
 llama-cpp.override { vulkanSupport = true; }