gh-116522: Stop the world before fork() and during shutdown (#116607)

This changes the free-threaded build to perform a stop-the-world pause
before deleting other thread states when forking and during shutdown.
This fixes some crashes when using multiprocessing and during shutdown
when running with `PYTHON_GIL=0`.

This also changes `PyOS_BeforeFork` to acquire the runtime lock
(i.e., `HEAD_LOCK(&_PyRuntime)`) before forking to ensure that data
protected by the runtime lock (and not just the GIL or stop-the-world)
is in a consistent state before forking.

Python/pylifecycle.c

@@ -1911,6 +1911,9 @@ Py_FinalizeEx(void)
     int malloc_stats = tstate->interp->config.malloc_stats;
 #endif
 
+    /* Ensure that remaining threads are detached */
+    _PyEval_StopTheWorldAll(runtime);
+
     /* Remaining daemon threads will automatically exit
        when they attempt to take the GIL (ex: PyEval_RestoreThread()). */
     _PyInterpreterState_SetFinalizing(tstate->interp, tstate);

Python/pystate.c

@@ -1692,6 +1692,10 @@ _PyThreadState_DeleteExcept(PyThreadState *tstate)
     PyInterpreterState *interp = tstate->interp;
     _PyRuntimeState *runtime = interp->runtime;
 
+#ifdef Py_GIL_DISABLED
+    assert(runtime->stoptheworld.world_stopped);
+#endif
+
     HEAD_LOCK(runtime);
     /* Remove all thread states, except tstate, from the linked list of
        thread states.  This will allow calling PyThreadState_Clear()
@@ -1710,6 +1714,8 @@ _PyThreadState_DeleteExcept(PyThreadState *tstate)
     interp->threads.head = tstate;
     HEAD_UNLOCK(runtime);
 
+    _PyEval_StartTheWorldAll(runtime);
+
     /* Clear and deallocate all stale thread states.  Even if this
        executes Python code, we should be safe since it executes
        in the current thread, not one of the stale threads. */