I'm not minded to build in extra components for the purpose of protection that isn't needed for the application in hand. I'm not switching motors and with a lead-acid battery a few hundred mA into the dew heaters at switch on isn't going to cause any woes.
I can find plenty of examples of no or much smaller gate resistors, and the transistors I've chosen are meant for direct drive.
Now I've wasted lots of brain power on trying to work pout what actually happens (I'm not blessed with SPICE).
With 100R a first estimate value for the peak current would be 5V/100R=50mA and a gate capacitance of 700pF means the time constant for the gate is 30nS, so after 30nS the current will be 24mA.
But the maximum current from the I/O pin also depends on the rise time and for an AVR that is about 20+(0.1 x load-capacitance-in-pF) or about 90nS.
Modelling it as best I can using excel, the gate voltage lags the output voltage and assuming the output voltage rises linearly, the maximum current through a 100R resistor is 15mA, around the middle of this graph.
, so the peak current would be limited more by the rise time of the gate than the resistor and is unlikely to approach the 40mA limit, even for nanoseconds.
Equally, pull down resistors are only an issue if the mosfets being switched on could cause problems – unlikely here.