Fix inconsistent speaker label handling in clip transcripts

[Copilot]

When generating clip transcripts, speaker labels were inconsistently applied: SRT-sourced text omitted the segment speaker label while fallback text included it.

Changes

• Modified transcript generation to always prepend segment speaker labels regardless of text source (SRT or fallback)
• Ensures consistent [Speaker]: text format across all transcript segments

// Before: speaker label only applied to fallback
if (relevantEntries.length > 0) {
  const text = relevantEntries.map(e => e.text).join(' ');
  return text;  // Missing speakerLabel
}
const text = segment.transcript || '';
return `${speakerLabel}${text}`;

// After: speaker label applied consistently
if (relevantEntries.length > 0) {
  const text = relevantEntries.map(e => e.text).join(' ');
  return `${speakerLabel}${text}`;
}
const text = segment.transcript || '';
return `${speakerLabel}${text}`;

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.