Replacing 4 together has the disadvantage that they are probably all from the same batch. It wouldn't be the first time that after a disk failed in RAID5 and was replaced, the next one gave up the ghost during the massive accesses for the reorganisation. Basically, all 1 SG and 3 WD run together, the only question is how the disk array is organised. For example, I have 2 NAS in 2 places that back up the same data independently of each other. 1 of them has only 1(!) of 4 slots occupied at the moment (it's brand new, only got RAID 1 a few months ago), the other 2 of 2 (RAID 1), whereby I regularly steal a disk from it and store it in a third place.

I would recommend the step-by-step version. Of the new 4 TB, only 2 TB would be used, but if over time the other three gradually fail and are each replaced with 4 TB capacity, you can expand the RAID.