I would only worry about inrush surge and possible ESD strikes. No need for huge ratings or anything, not even an SMAJ should be needed I think.
Inrush surge is mainly a current thing, but it can become quite substantial voltage when combined with the nonlinear dielectric of type 2 capacitors (C5, C6).
The linear case, with no resistance and zero initial conditions, is an inrush current equal to Vin / sqrt(L / C), and a peak voltage of 2Vin; where L is the loop inductance (mostly the cable, ca. 0.5uH per m length), and C is the total input capacitance. The voltage peak is somewhat lower with ESR of course, and further ringing is dampened.
The nonlinear case, where C decreases with increasing V, allows the peak to rise much higher, often over 3Vin. How much, depends on the C(V) curve of your chosen parts; you'll have to go digging for the char sheets, if they provide them at all. Also, hopefully you'll check for reasonably nominal value at operating voltage; 1uF 0603s being say -50% at 5V (regardless of voltage rating) is not uncommon. 0805+ should be fine, but do check as they can always skimp on the internals, the only thing you're really guaranteed is a too-small part won't hold enough charge.
So, for these cases, having a TVS to clamp the peak, and an LDO rated for the clamped peak, is a good idea. Typically a TVS rated for nominal operating voltage (i.e. 5V) will clamp around 30% higher, or maybe worse I guess at lower voltages (I'd expect 8V here?). So a 10V rating for those parts should be adequate.
Tim