What's the current state of digitalWrite (slowness) ?

We all remember from before, that DigitalWrite were pretty slow before, - so people came up with digitalFastWrite and/or libraries like DigitalIO

Some used more direct, C-like commands instead of digitalWrite.

My question is, are all these "challenges" solved today ? - on 328 based controllers as well ? Is there any significant difference (for time critical applications, or that will free some processing power) between the digitalWrite and the mantioned methods ?

They are not challenges they are how things are. The faster a digital write the less things it can do, like checking if the PWM is on.
The fastest way to access pins is with direct port manapulation if your skill is up to it.

yep, I just saw that in the source - thank you.