make llama server configurable

src/solanaceae/llama-cpp-web/llama_cpp_web_impl.hpp

@@ -2,6 +2,8 @@
 
 #include "./text_completion_interface.hpp"
 
+#include <solanaceae/util/config_model.hpp>
+
 #include <httplib.h>
 #include <nlohmann/json_fwd.hpp>
 
@@ -9,14 +11,19 @@
 #include <atomic>
 
 struct LlamaCppWeb : public TextCompletionI {
-	// this mutex locks internally
-	httplib::Client _cli{"http://localhost:8080"};
+	ConfigModelI& _conf;
+
+	// this mutex-locks internally
+	httplib::Client _cli;
 
 	// this is a bad idea
 	static std::minstd_rand thread_local _rng;
 
 	std::atomic<bool> _use_server_cache {true};
 
+	LlamaCppWeb(
+		ConfigModelI& conf
+	);
 	~LlamaCppWeb(void);
 
 	bool isGood(void) override;