Build speaker-labeled Sapat transcription reviews

[BerqiaMouad]

Summary

• Add a long-form Daytona guide for building a speaker-labeled transcript review pipeline around Sapat.
• Ground the workflow in current Sapat behavior: ffmpeg MP3 conversion, .mp4 file/directory processing, OpenAI/Groq/Azure provider selection, quality/language/prompt/temperature flags, optional correction, and sidecar .txt transcript output.
• Add a speaker-labeled transcript definition, first-time author profile, and original PNG workflow diagram.
• Keep the scope distinct from recent Sapat submissions by focusing on speaker maps, manual review packets, uncertainty markers, and promotion gates before downstream AI use.

Validation

• npx --yes markdownlint-cli guides/20260513_speaker_labeled_sapat_transcription_review_pipeline.md definitions/20260513_definition_speaker_labeled_transcript.md authors/berqia_mouad.md --ignore node_modules
• git diff --check
• File naming compliance check matching .github/workflows/content-check.yml for changed guide and asset paths

/claim #13

Transparency: AI-assisted with Codex; I checked the Sapat README and source before writing the commands and review workflow.