frank-primary: patch is Dockerfile-only on feat/ds41-engine

The llama.cpp hunk (ggml_cuda_node_launches_nothing) was needed on main's
tree; the branch restructured that file and compiles clean, so only the
apt-list and cmake-3.31 hunks remain. Regenerated from the host clone.

dankserver/frank-primary/lucebox-gfx1201.patch

@@ -38,31 +38,3 @@ index 4ea7a1d7..0e22355b 100644
      && rm -rf /var/lib/apt/lists/*
  
  # uv (pinned) manages Python 3.12 + resolves the luce + pflash
-diff --git a/server/deps/llama.cpp/ggml/src/ggml-cuda/ggml-cuda.cu b/server/deps/llama.cpp/ggml/src/ggml-cuda/ggml-cuda.cu
-index 1e6ca455..5f4b3035 100644
---- a/server/deps/llama.cpp/ggml/src/ggml-cuda/ggml-cuda.cu
-+++ b/server/deps/llama.cpp/ggml/src/ggml-cuda/ggml-cuda.cu
-@@ -3916,6 +3916,12 @@ static void ggml_backend_cuda_synchronize(ggml_backend_t backend) {
-     GGML_UNUSED(backend);
- }
- 
-+// (moved above the USE_CUDA_GRAPH guard: ggml_cuda_node_is_noop
-+// calls this unconditionally)
-+static bool ggml_cuda_node_launches_nothing(const ggml_tensor * node) {
-+    return ggml_is_empty(node) || node->op == GGML_OP_RESHAPE || node->op == GGML_OP_TRANSPOSE ||
-+        node->op == GGML_OP_VIEW || node->op == GGML_OP_PERMUTE || node->op == GGML_OP_NONE;
-+}
- #ifdef USE_CUDA_GRAPH
- 
- // GGML_CUDA_COLLECTIVE_GRAPH_CAPTURE=1: allow graph capture of a graph that
-@@ -3930,10 +3936,6 @@ static bool ggml_cuda_cluster_capture_allowed() {
- }
- 
- // View-like and empty nodes: the evaluation loops launch nothing for them.
--static bool ggml_cuda_node_launches_nothing(const ggml_tensor * node) {
--    return ggml_is_empty(node) || node->op == GGML_OP_RESHAPE || node->op == GGML_OP_TRANSPOSE ||
--        node->op == GGML_OP_VIEW || node->op == GGML_OP_PERMUTE || node->op == GGML_OP_NONE;
--}
- 
- static bool ggml_cuda_graph_check_compability(ggml_cgraph * cgraph) {
-