Make more ML improvements

- Fix UX issues with llama.com
- Do housekeeping on libm code
- Add more vectorization to GGML
- Get GGJT quantizer programs working well
- Have the quantizer keep the output layer as f16c
- Prefetching improves performance 15% if you use fewer threads

libc/calls/getcpucount.c

@@ -95,6 +95,6 @@ __attribute__((__constructor__)) static void _getcpucount_init(void) {
  *
  * @return cpu count or 0 if it couldn't be determined
  */
-unsigned _getcpucount(void) {
+int _getcpucount(void) {
   return g_cpucount;
 }