jobd.pl: change default number of parallel jobs
The limiting factor tends not to be CPU when the number of CPU cores
available increases, but rather the random access speed of the hard
disk. Having a large number of parallel jobs can actually reduce
performance due to hard-disk thrashing.
Nevertheless, there is a period of just waiting for network activity
while doing mirror updates that does not cause disk thrashing at all.
Therefore scale back on the default number of parallel jobs as the
number of detected CPU cores increases and cap the maximum default
to no more than sixteen.
Of course any explicitly set value will continue to be used, this
change only affects the default if no value is specified explicitly.
Signed-off-by: Kyle J. McKay <mackyle@gmail.com>