fix: pass user_name to all transcript callers (not just summaries)

[beastoin]

Problem

Conversation transcripts show generic "User: ..." instead of the user's actual configured name (e.g. "Stephen: ...") across multiple code paths. The user_name parameter exists in segments_as_string() but most callers don't pass it.

PR #5369 by @bzanghi correctly fixes the two highest-impact paths (_get_structured and _trigger_apps), but 4 other callers of get_transcript() and 6 direct callers of segments_as_string() still use the generic "User" label.

Affected paths (not covered by #5369)

get_transcript() callers still missing user_name:

File	Function	Impact
backend/utils/llm/trends.py:39	Trends extraction	Low — analytical feature
backend/utils/llm/chat.py:865	RAG context retrieval	Medium — affects chat context
backend/utils/llm/chat.py:899	Emotional message generation	Medium — daily summaries
backend/models/conversation.py:397	conversations_to_string()	Low — bulk display

segments_as_string() direct callers still missing user_name:

File	Line	Function
backend/utils/llm/followup.py	9	Follow-up question generation
backend/utils/llm/chat.py	216	Chunk extraction for RAG
backend/utils/llm/chat.py	1331	Question generation
backend/utils/llm/chat.py	1352	Emotional context retrieval
backend/utils/retrieval/rag.py	57	Conversation chunk retrieval
backend/scripts/rag/app.py	219	RAG app script

Suggested fix

Thread user_name through all callers, not just the summarization paths. The pattern from #5369 works well — fetch get_user_name(uid, use_default=False) once per request and pass it down.

For segments_as_string() direct callers that don't have access to uid, consider adding user_name as a parameter to the calling function signatures (same approach as #5369).

Related

• PR fix: pass user's configured name to transcript formatter for summary … #5369 — partial fix (summaries + apps only)