Oh, that would be for readability.
if you use a separate process for each signal, the code becomes cluttered and it becomes hard to determine which signals are conceptually related to other signals. On the other side, if you assign unrelated signals within a process, it makes these signals appear conceptually related to the rest of the signals in the process. In more extreme cases, like when only one process is used in a file, it can make it hard to determine which signals are conceptually related.
Some people take this idea a bit further, and use blocks, or comments, to denote groups of related logic -- processes, assignments, components, ect...