The diode work on the assumption that turning power off means that Vcc gets connected to GND
Well, yes. If power stays at 3.3V, it's not off, and you don't need another reset.
Systems that don't reset properly on a power glitch are a nightmare - which is why most micros have proper brownout detectors, which will work better than the diode hack.
Reset supervisor chips come in SOT23 (so, since you could ditch at least the capacitor, and probably the resistor, they'd be smaller than the diode hack), and have accurate trip points and reset delays. I use them all over the place. They're cheap, reliable, available, small. All sorts of good things! They're also low power, so if you want to bodge a reset button in, you can usually just feed the reset chip through a resistor, and short across its power pins with a switch to start a reset cycle, so you don't need to stock chips with a reset-in pin for when you're prototyping.
Using power supplies' power-good outputs is often also a good way of making a reset, if they come for free.
>The reset diode is the AVR042 suggestion for ESD protection.
Hmm. That sounds unlikely. Protection for the AVR's ESD protection structure, maybe.