Speed up Acquisition

[drowe67]

Goal: reduce CPU load of acquisition in search state

We can probably subsample the time/freq grid and still get acceptable results.
Need to verify based on ctests, possibly code up new ctests if not comprehensive enough
Maybe able to use FFTs to speed correlation computations
Refs: radae.pdf, acquisition code in dsp.py and rx.py
Skills: Python, numpy, basic DSP, e.g. you can understand dot product and exp(-j \omega n)