* (experiment) measure raw signal for radio3, radio4 for ~ 1 minute
   * different gain settings
   * with/without RTL AGC mode
   * with several IF gain settings
* confirm theories about effect of gain, IF gain, AGC
* look for effect of gain on baseband SNR
* look for effect of ADC calibration on baseband SNR
* look for effect of IF bandwidth on baseband SNR

* (experiment) try if RTL2832 FIR filter can be optimized
* (feature) support 'M' 'k' suffixes for sample rates and tuning frequency
* (feature) implement off-line FM decoder in Python for experimentation
* (feature) implement stereo pilot pulse-per-second
* (quality) consider DC offset calibration
* (speedup) maybe replace high-order FIR downsampling filter with 2nd order butterworth followed by lower order FIR filter
* figure out why we sometimes lose stereo lock
* it looks like IF level sometimes varies so much that it saturates the receiver; perhaps this can be solved by dynamically managing the hardware gain in response to level measurements
* investigate problem at low IF sample rates; it seems that librtlsdr is not telling the truth about the actual sample rate.
* (quality) figure out if I/Q balance can improve weak stations
* (quality) figure out if hardware gain settings can improve weak stations
* (feature) implement RDS decoding
* (quality) consider FM demodulation with PLL instead of phase discriminator