Support longer input files by splitting them server-side#23

The current API restricts files over ~60m, and recommends that the user split them (https://docs.gladia.io/reference/limitations-and-best-practices#splitting-oversize-audio-files). It’d be nice if the server could instead do that internally, since it’s likely that you’ll do a better job of the splitting (especially in batch use-cases) than the user. For example, you could run transcription on the first 60m of the audio file, find the end time code of the second-to-last caption, run transcription from that point + 60m, and repeat until the long video is all transcoded. At that point, you could stitch together the caption outputs while eliminating the overlapping couple of captions, and return it all as a single transcription. This could be implemented client-side, but it’d be oh-so-convenient if the server could just take care of it. Especially when providing a URL (like YouTube( rather than a local audio file!