Add pipeline guide and enhance CMake configuration for llama integration

pipeline/CMakeLists.txt

@@ -19,6 +19,23 @@ FetchContent_Declare(
 )
 FetchContent_MakeAvailable(nlohmann_json)
 
+FetchContent_Declare(
+    llama
+    GIT_REPOSITORY https://github.com/ggml-org/llama.cpp.git
+    # Stable release tag: b8485 (commit 31a5cf4c3f5d3af7f16fc4abc9baa75f8d568421)
+    GIT_TAG        31a5cf4c3f5d3af7f16fc4abc9baa75f8d568421
+)
+FetchContent_MakeAvailable(llama)
+
+# Workaround for upstream llama.cpp release stream (b8485/b8496) missing
+# <algorithm> include in llama-quant.cpp where std::sort is used.
+# Remove once fixed upstream.
+if(TARGET llama)
+    target_compile_options(llama PRIVATE
+        $<$<COMPILE_LANGUAGE:CXX>:-include algorithm>
+    )
+endif()
+
 file(GLOB_RECURSE SOURCES CONFIGURE_DEPENDS
     src/*.cpp
     src/*.h
@@ -36,6 +53,7 @@ target_link_libraries(biergarten-pipeline
         CURL::libcurl
         nlohmann_json::nlohmann_json
         Boost::unit_test_framework
+        llama
 )
 
 target_compile_options(biergarten-pipeline PRIVATE
@@ -95,6 +113,7 @@ if(BUILD_TESTING)
             Boost::unit_test_framework
             CURL::libcurl
             nlohmann_json::nlohmann_json
+            llama
       )
 
       add_test(