Switched to using complex_to_mag_squared vs. mag. Based on an observation that

the optimal slicer threshold is 0.5(V**2)t. Less CPU load, better performance.
Some configurations of threshold and --pmf result in slightly worse
performance. Most result in better performance.

python/rx_path.py

@@ -35,7 +35,7 @@ class rx_path(gr.hier_block2):
         self._spc = int(rate/2e6)
 
         # Convert incoming I/Q baseband to amplitude
-        self._demod = blocks.complex_to_mag()
+        self._demod = blocks.complex_to_mag_squared()
         self._bb = self._demod
 
         # Pulse matched filter for 0.5us pulses