drivers: clock_control: Enable LFSTARTED event for nrf5

The LFSTARTED event was disabled so it was impossible to wake the
CPU up on LF clock being ready. The Bluetooth stack was putting
the CPU to sleep, expecting to be woken up on LFSTARTED event.
As the event never triggered, the CPU was woken up seconds later
by a different event introducing a multi-second startup delay.

Bug introduced by 23c92100.

Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>