As far as I remember, you need to hit the resonant frequency of a particular hair to trigger a "sound" response, so frequencies higher than 20KHz might excite them, but if you're not getting resonance, nothing triggers.

No this is wrong. Each hair is excited by the amount of its particular resonant frequeny in the sound hitting it. If a violin note is heard, that note only has a few discrete frequencies in it and so a few hairs are very excited about it and the brain (of the trained violinist with perfect pitch anyway) goes "oh, A 440." If white noise loud enough to hear is hitting the ear, then essentially all the hairs are excited because all frequencies are present in white noise, and the brain goes "sounds like the ocean."

As to excitement by sound above 20 kHz, a very high frequency ultrasound, say at 100 kHz, can be modulated with the vibrations associated with a violin string, much as sound can be modulated on radio carriers. Such ultrasound hitting a human ear can actually cause the appropriate hairs to be excited so that the brain goes "oh, A 440." The phenomenon relies on the non-linear response of cochlear hairs and highly directional speakers based on this effect have been built and demonstrated. See for example http://www.holosonics.com/

That's a somewhat crude way of putting it; when studying a resonator it's better to look at the q factor: https://en.wikipedia.org/wiki/Q_factor

Lower q factor means a higher spread of frequencies can trigger them. Mammalian hair cells have q factors of 5-10. Q=10 is pretty high for a biological resonator, but pretty low compared to, say, even crude electronic equipment. A typical LC oscillator has Q of 100 or more.