If you think about what the PC is modeled after. That is a channel strip on a mixing desk, it sort of makes sense. Compression was used prior to the strip itself to control the dynamics of the signal. EQ would be used to cut out a space for other instruments, and so on.
I like compression as the last thing in the chain because I don't want the signal to clip. The use of the CA 2A is a different case and I like it as the first module.
There is no law saying you can't have what you want where you want it, though.