A number of databases (e.g. Oracle) allow you to specify the size in bytes or characters e.g., "a_col varchar(128 char)" so encoding does not come in to it.

varchar(255) is probably seen a lot simply because in many databases it was the largest varchar you could have. Even now many databases only support varchars of 2000 or 4000.

In my opinion, it is mostly cargo cult but I'll wait for someone to come up with a reason. I do it myself quite a bit when needing to hold an error message in a varchar. I know it is not going to be bigger than 255 but I'm too lazy to work out if it could be smaller.