Version 4.0.0

• RTLSDR-Airband is now built with CMake. Refer to the wiki for updated
  compilation instructions.
• When compiling the program, a new PLATFORM value native can now be
  specified. It enables -march=native -mtune=native compilation options. This
  causes the compiler to apply the most appropriate optimizations for the
  hardware on which the app is being built (thx @charlie-foxtrot).
• BACKWARDS-INCOMPATIBLE CHANGE: Signal level and noise level estimates
  displayed in the textual waterfalls are now expressed in dBFS (decibels
  related to the full scale of the analog-to-digital converter). The main
  benefit of the new approach is that these values do not depend on the
  fft_size value(thx @charlie-foxtrot).
• BACKWARDS-INCOMPATIBLE CHANGE: Improved squelch algorithm with new
  configuration parameters. squelch keyword has been replaced with
  squelch_threshold which takes an absolute signal value in dBFS as an
  argument. Alternatively, a minimum signal-to-noise ratio (in dB) that should
  trigger the squelch might be configured using squelch_snr_threshold option
  (thx @charlie-foxtrot).
• BACKWARDS-INCOMPATIBLE CHANGE: include_freq config option for file outputs
  now causes the frequency to be appended after the timestamp rather than
  before it. This feature now works correctly in scan mode, when
  split_on_transmission feature is enabled. (thx @charlie-foxtrot).
• BACKWARDS-INCOMPATIBLE CHANGE: sample format in files produced by rawfile
  outputs has been changed from CS16 to CF32. File name suffix is now .cf32.
• Improved squelch indicator in the textual waterfalls. In addition to the *
  character indicating that the squelch is open, there is also a ~ character
  indicating that the channel has a signal that is being suppressed because it
  is outside the band of the channel filter (thx @charlie-foxtrot).
• New output type udp_stream for sending uncompressed audio to another host
  via UDP/IP (thx @charlie-foxtrot).
• Added multiple_output_threads global option. When set to true, a separate
  output thread is spawned for each device (thx @charlie-foxtrot).
• Modulation in scan mode is now configurable per channel (thx
  @charlie-foxtrot).
• SoapySDR errors like TIMEOUT or OVERFLOW are no longer treated as fatal. They
  often appear intermittently, especially when the CPU usage is high. There is
  no point in failing the input in this case.
• Added .tmp suffix to the names of the output files currently being written
  to. The suffix is removed when the file is closed. External applications that
  consume recorded files can now figure out which files are not yet complete.
• Added logging and statistics for output thread overruns and mixer
  input/output overruns (thx @charlie-foxtrot).
• The program can now be built on MacOS.
• Miscellaneous bug fixes and code cleanups.