Benchmark SAPAT Transcription Providers in Daytona

[rohitmulani63-ops]

/claim #13
fix #13

Summary

• Add a focused Daytona guide for benchmarking SAPAT transcription providers in a repeatable workspace.
• Cover ffmpeg setup, Python dependencies, provider configuration, separate OpenAI/Groq/Azure runs, transcript preservation, and scorecard-based review.
• Add an original SVG workflow diagram, a speech-to-text definition file, and a first-time author profile.

Why this is distinct

This is not another generic SAPAT setup guide. It shows AI engineers how to compare the current SAPAT CLI providers against the same recording, same prompt, and same quality setting, then choose based on terminology accuracy, cleanup effort, runtime notes, and operational fit.

Notes

• Documentation-only change.
• The guide uses the current SAPAT CLI choices: openai, groq, and azure.
• The guide distinguishes active CLI providers from future provider placeholders in .env.example.
• No secrets, API keys, private videos, generated transcripts, payout details, or private data are included.

Validation

• Local word count check: 1,752 words.
• git diff --check passed.
• SVG XML parse passed.
• Targeted markdownlint passed for the guide, definition, and author profile.
• Local secret-pattern scan: 0 obvious token patterns found.