Hi,
the reason is simple: if clock of CMU fails, the watchdog resets application (because is running on another clock) - assuming that CMU's reset is the safe state. Next actions depend on supervisor system.
If CMU and watchdog share the same clock, the application got stuck during clock malfunction.
Best regards,
Pavel