One reason a fast power-on is needed is to guarantee that the oscillator gets the needed kick to start. There are oscillators not showing enough start-up reserve, which you can block with a very slowly power-on.

One reason to have a stable reset signal from Vcc=0V on is, that micros start several internal routines at power-on, which eventually are invoked at different levels of supply voltage. Start-up of oscillator is one such mechanism. The idea of the reset signal is, to start certain internal "affairs" at the same moment and by this to synchronize them.

Because of that, a sufficiently fast power-up is needed (to allow the oscillator to start), a proper reset signal is needed from Vcc=0V on (to keep all internal affairs quiet), and a suffieciently long reset active time is needed (to guarantee a synchronisation of certain internal affairs).

When flash micros came into play, which can be programmed by the invoke of internal ISP routines (another sort of internal affair) an improper reset not only makes your product a toy, but can also result in a damage of flash code memory content...
Especially here another feature of reset circuitry is needed, brown-out detection.

What are these internal affairs?
I don't exactly know. But I remember, that certain micros had internal charge pumps to bring the substrate on a certain potential. Also, don't forget that certain micros are not statically working but need a minimum clock frequency. This should provide ample food for thought...

Kai