Unify benchmarks from whisperkittools

[EduardoPach]

What does this PR do?

This PR serves as the first step to unify the benchmarks once done in whisperkittools to SDBench by adding some pipelines that were only in whisperkittools, and adding support for the main datasets used in whisperkittools

• Adds WhisperKit and Apple's SpeechAnalyzer as pipelines
• Adds useful aliases for new pipelines
• Add useful aliases for new datasets common-voice, librispeech, and earnings22

[atiorh]

Looks great - 1 NIT comment

[arda-argmax]

LGTM! ✅

Suggestions for future improvement:

• We can add support for additional pipelines available in whisperkittools, such as WhisperCPP and WhisperMLX.
• It might also be helpful to enable support for local/custom models within the WhisperKitTranscriptionPipeline.