Complete waveform loudness data generated when loading file
Part of: #83
Currently, decibels generates the loudness data for a complete audio file as fast as possible after playing it.
For very long audio files (ex. podcasts), this can keep a single CPU core busy for multiple minutes, especially on embedded devices.
A solution could be to incrementally load the waveform data at the same pace as the file is played. This could still be problematic when seeking inside the audio file, but should be fairly straightforward to implement.
A more sophisticated solution could be to load the waveform data in fixed-sized chunks (or generally "on-demand"), but this would probably require a whole refactor of the waveform visualization code.