This is a backwards-incompatible change that I knew I'd have to
introduce sooner or later. Which means now.
Long story short: when I first started writing this library, I wanted to
make things simpler by treating the entire talker+sentence as one big
ID. Of course, this is an oversimplification, as there are different
talkers out there.
Turns out they're more common than I initially thought. I just came
across a module (read: I had one soldered to my board by our crazy
hardware guy) that spits out GPRMC, GLRMC or GNRMC depending on where it
got the data. I could have kludged around this, but here's this change,
for the purity of code.
Fortunately, we're the only users of this codebase as far as I know, so
no real harm is done - but it's a lesson to design my interfaces right
from the very start so I don't have to break compatibility and write
apologetic messages later on.
