fix(turboquant): repair pinned WHT declaration

The known-good pin still carries a second definition of the WHT group-size symbol, which breaks static backend builds. Carry the declaration-only form while this older upstream commit is pinned.

Assisted-by: Codex:gpt-5 [Codex]

Author: localai-org-maint-bot

backend/cpp/turboquant/patches/0002-fix-static-wht-group-size-declaration.patch

@@ -0,0 +1,14 @@
+ggml-cpu: keep the WHT group-size symbol declaration-only
+
+The pinned fork commit declares turbo3_cpu_wht_group_size without extern in
+ops.cpp, creating a second definition in libggml-cpu. Keep the single definition
+from ggml-turbo-quant.c so static backend builds can link the two libraries.
+
+diff --git a/ggml/src/ggml-cpu/ops.cpp b/ggml/src/ggml-cpu/ops.cpp
+index 35f7cf1de..b3ca80d79 100644
+--- a/ggml/src/ggml-cpu/ops.cpp
++++ b/ggml/src/ggml-cpu/ops.cpp
+@@ -16,2 +16,2 @@ extern "C" {
+-GGML_API int turbo3_cpu_wht_group_size;
++extern int turbo3_cpu_wht_group_size;
+ }