Fix #274: Avoid possible deadlock of timer callback

If the tick_time from the wait routine was perpetually
greater than the interval time, eventually the wait_time
became such that it was always below zero.  Once this
occurs, application callbacks would cease entirely.

To avoid this, run the callback condition in a loop
for periodic timers only (not for one-shot config).

src/os/shared/osapi-timebase.c

@@ -494,7 +494,7 @@ void OS_TimeBase_CallbackThread(uint32 timebase_id)
                 timecb = &OS_timecb_table[curr_cb_local_id];
                 saved_wait_time = timecb->wait_time;
                 timecb->wait_time -= tick_time;
-                if (timecb->wait_time <= 0)
+                while (timecb->wait_time <= 0)
                 {
                     timecb->wait_time += timecb->interval_time;
 
@@ -519,6 +519,14 @@ void OS_TimeBase_CallbackThread(uint32 timebase_id)
                     {
                         (*timecb->callback_ptr)(curr_cb_public_id, timecb->callback_arg);
                     }
+
+                    /*
+                     * Do not repeat the loop unless interval_time is configured.
+                     */
+                    if (timecb->interval_time <= 0)
+                    {
+                        break;
+                    }
                 }
                 curr_cb_local_id = timecb->next_ref;
             }