Add a mechanism to reset the default task spawn order

Add `qthread_reset_target_shep()`, which resets the shepherd that tasks
created with `NO_SHEPHERD` get spawned to. This is useful in Chapel for
cases where we want to use the default spawn location, but get good NUMA
affinity between data parallel loops sepreated by some other task spawn.

Just using the `qthread_fork_to` API would disable future stealing, so
this is a mechanism  to get tasks in consecutive data parallel loops to
spawn to the same shepherds without preventing stealing.

This was originally done in chapel-lang/chapel 12868, which includes
some perf results. At the time I expected we'd update our shim to use
`qthread_fork_to`, but had not considered the work-stealing implications
at the time.

include/qthread/qthread.h

@@ -333,6 +333,9 @@ int qthread_spawn(qthread_f             f,
 /* This is a function to move a thread from one shepherd to another. */
 int qthread_migrate_to(const qthread_shepherd_id_t shepherd);
 
+/* Resets the default shepherd spawn order for tasks that use NO_SHEPHERD */
+void qthread_reset_target_shep(void);
+
 /* This function sets the debug level if debugging has been enabled */
 int qthread_debuglevel(int);
 

src/qthread.c

@@ -3000,6 +3000,19 @@ int API_FUNC qthread_migrate_to(const qthread_shepherd_id_t shepherd)
     }
 }                      /*}}} */
 
+void API_FUNC qthread_reset_target_shep(void) {
+    assert(qthread_library_initialized);
+    qthread_t *me = qthread_internal_self();
+
+    if (me) {
+        assert(me->rdata);
+        me->rdata->shepherd_ptr->sched_shepherd = 0;
+    } else {
+        qlib->sched_shepherd = 0;
+        MACHINE_FENCE;
+    }
+}
+
 
 /* These are just accessor functions */
 unsigned int API_FUNC qthread_id(void)