[raudio] WaveCrop() is broken

[ListeriaM]

initSample is treated as being in both [frames] and [frames * channels] units. This was probably an oversight in 7f6cd93 as initSample is still multiplied by wave->channels in the call to memcpy(3) which renders the function unusable.

Although I think it would make more sense for initSample and finalSample to be measured in [frames] (and renamed to initFrame and endFrame), as I can't think of a reason to support splitting frames, because with interleaved channel data, you may end up getting different starting and ending offsets for different channels, and the channels' data would be shifted, i.e: the n'th channel would have the data for the k'th channel, with k = (n + initSamples) % wave->channels.

I can make a PR for this if needed, but unless there's a good reason for using samples, I'd go ahead and use frames instead.

[raysan5]

@ListeriaM Good catch! Feel free to send a PR and go for frames!