fix: synchronize lazy mutex initialization (closes #297)

Two threads locking the same mutex for the first time could both run the
unsynchronized lazy init, corrupting the underlying pthread mutex and aborting
or deadlocking. Build the object and publish it with a single atomic
compare-and-swap; threads that lose the race free the object they built. This
needs no statically-initializable guard, so it stays valid on Windows 2000.

src/htsthread.c

@@ -193,7 +193,23 @@ HTSEXT_API void hts_mutexfree(htsmutex * mutex) {
 HTSEXT_API void hts_mutexlock(htsmutex * mutex) {
   assertf(mutex != NULL);
   if (*mutex == HTSMUTEX_INIT) {        /* must be initialized */
-    hts_mutexinit(mutex);
+    /* Initialize exactly once, even when several threads race to lock the same
+       mutex for the first time. Build our own object, then publish it with a
+       single atomic compare-and-swap; the threads that lose the race free the
+       object they built (issue #297). No static guard is needed, which keeps
+       this safe on Windows 2000 (no statically-initializable lock there). */
+    htsmutex created = HTSMUTEX_INIT;
+
+    hts_mutexinit(&created);
+#ifdef _WIN32
+    if (InterlockedCompareExchangePointer((PVOID volatile *) mutex, created,
+                                          HTSMUTEX_INIT) != HTSMUTEX_INIT)
+#else
+    if (!__sync_bool_compare_and_swap(mutex, HTSMUTEX_INIT, created))
+#endif
+    {
+      hts_mutexfree(&created);
+    }
   }
   assertf(*mutex != NULL);
 #ifdef _WIN32