I mean, there are digital servos that are much better, and need I2C or SPI to control, that's another story.
But analog servos could easily change the PWM range to span e.g. 1% to 99% instead of the silly 8%-10% or whatever it is now, and software would be trivial to change, the PWM pins on microcontrollers can do any duty cycle. It would be an easy 10X resolution improvement because the PWM settings on these microcontrollers are usually 8-12 bits for the entire 0%-100% range.
Maybe, but they already have PWM controllers in them, it's just they don't use the full range so you lose a few bits. By just using the full range it would be incredibly trivial change in software to just use the full PWM range again.
Maybe, if the controllers are internally digital. If they're internally analog requiring their deviation from linearity to be insignificant over the 1%-99% range of PWM values rather than the 8%-10% range is likely to require extra complexity.
Now that Padauk has microcontrollers that cost less than most discrete transistors it might seem like an obvious improvement to redesign such controllers to be internally digital --- but Padauk's microcontrollers were out of stock this year, and discrete transistors weren't.
There are slightly larger servos which are much better, all the components are better, but they're expensive ($30 or so each) and beyond that, there are robotics servos (like dynamixels) but those are fairly pricey.
