Now that we don't support ARMv3 anymore, the loop based delay code can
convert microsecs into number of loops using a 64-bit multiplication
and more precision.

This allows us to lift the hard limit of 3355 on the bogomips value as
loops_per_jiffy may now safely span the full 32-bit range.
Signed-off-by: Nicolas Pitre <nico@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>