OK, after some pacnew-chaser I come to conclusion:
Pacnew files cannot be easily determined as better in overall, because of various situations:
- confs often contain important settings for current system and pacnew are just blank, for example pssdw or groups file, that contains just root entry and none of the system users/groups, it cannot be and shouldn't be merged in any circumstances
- sometimes we used personal settings, like setting an unstable or testing branch for pacman-mirrors so default pacnew file doesn't reflect that
- some pacnew files indeed have new options that come from new kernel features like with TLP, so it's important to update those and yet don't update personalized blacklisted devices and some other non-standard settings that were changed (often not directly in conf but by commands or different system tools)
- some pacnew have just updated hashed descriptions so merge all the way
- some pacnew lack of description and are in more raw form then the current ones, so it's good to keep old ones hashed lines for better understanding what is happening in given conf
- and some more various sitiuations
It takes lot of time and sometimes research to asses whether I should change something or not and documentation is not always clear enough. Still, such tools as pacnew-chaser and meld help tremendously in that task. Previously I compared files by opening both of them, snapping them to the sides (left and right) on another virtual desktop and look trough. As you can imagine, this is troublesome and not well effective.
However, I found some issues with pacnew-chaser. Sometimes after editing files in meld, pacnew-chaser seems to be frozen and unresponsive and I have to shut it down forcefully. It happens pretty often. Will investigate what is happening.