Rationale behind using the scaled coordinate/how to handle out-of-box values in post-processing

Dear all,
I have a two elementary questions about the dump atom and similar dump commands:
1- What is the rationale behind outputting the scaled atoms coordinates as the default setting for the dump atom command? Is it computationally better?
2- If the scaled or unsealed coordinates are out of the simulation box size, how I need to fix them in the data analyzing step (I know there is a dump_modify pbc on/off command but it is computationally expensive in the case of my simulations)?
Thanks in advance for your help,
Amir Sadeghi