There is no maximum pulse width for the RESET* pin on the 9914. RESET*
just resets most of the internal registers when asserted, so it doesn't
need to be asserted for very long after VCC comes up. I don't have an
exact number for the minimum but a few microseconds should be
sufficient. Typically there is some sort of power monitor circuit that
will keep a reset asserted for several milliseconds or longer after VCC
is stable.
There is also no spec for the control lines being correctly driven
prior to RESET* unasserting but the minimum required time is very
short.
As for the reverse current, is the device connected to the 9914 driving
the interface signals to the 9914 when the 9914 is not powered? Driving
signals to the 9914 while it is not powered may cause damage. If the
device connected to the 9914 is powered on before the 9914, is it
possible to tri-state the control signals and place pullups/pulldowns
to the same power rail that powers the 9914?