I've been observing random failures in gdb.base/freebpcmd.exp on a 
bare-metal target using RSP with a slow debug probe.  It seems that the 
hardwired timeout of 120 seconds for this test is just a little too low 
for this target.  It seems better to use with_timeout_factor instead of 
a hardwired value; choosing a factor of 10 here is totally random, but 
it seems to be plenty on this target (its default timeout is 60).

OK to commit?

-Sandra