Make _Thread_local work across platforms

We now rewrite the binary image at runtime on Windows and XNU to change
mov %fs:0,%reg instructions to use %gs instead. There's also simpler
threading API introduced by this change and it's called _spawn() and
_join(), which has replaced most clone() usage.

libc/str/lz4decode.c

@@ -32,7 +32,7 @@
  * they are passed in the ≤64kb bytes preceding src.
  *
  * @return pointer to end of decoded data, similar to mempcpy()
- * @see mapanon(), lz4check()
+ * @see _mapanon(), lz4check()
  */
 void *lz4decode(void *dest, const void *src) {
   const unsigned char *frame, *block;