Research Timeline Rendering

The system SHALL render research results in channel-specific section order:

PDF Report: Title → Summary → References → Cited Paragraphs → Research Timeline
Email Body: Title → Summary → References → Cited Paragraphs → Workflow Signature
Chat UI: Summary → References → Cited Paragraphs → [Collapsible Timeline]

The system SHALL NOT render the Research Timeline or Cited Paragraphs section between the Summary and References sections.

The system SHALL generate a contextual title for PDF research reports using AI, rather than relying on the email subject line or first sentence of the query.

The generated title SHALL be:
- 3-8 words in length
- Descriptive of the research question or topic
- Used as the PDF report heading
- Used as the PDF filename (with appropriate sanitization)

Chat sidebar titles SHALL remain unchanged (derived from first words of message).

The system SHALL render a compact workflow signature in email body instead of the full Research Timeline.

The workflow signature SHALL include:
- Completion indicator: ✅ Completed workflow: research_pipeline
- Processing metadata: processing time, tokens used
- Completed todos list with summary stats (e.g., "3 web searches", "read 1 Drive file")
- Pointer to attachment: 📎 Full research timeline available in attached PDF report

The system SHALL render a References section as a numbered bibliography containing:
- Sequential numbering [1], [2], etc. matching inline citations
- Source title/name only (short form)

The References section SHALL appear before the Cited Paragraphs section.

The system SHALL render a Cited Paragraphs section (🔍) containing full source details:
- Reference number [N] matching References section
- Source title
- Source type indicator (Web Result #N, Google Drive File, SharePoint File, etc.)
- Text snippet/excerpt from the source
- URL or file identifier

The Cited Paragraphs section SHALL appear after the References section.

The system SHALL display each cited insight in the following format:
1. Insight text (paraphrased finding) followed by citation number(s) in brackets [N]
2. Supporting quote(s) displayed in *italics* to differentiate from paraphrased text

The system SHALL NOT display confidence scores to end users.

The system SHALL serialize complete CitedInsight data when sending research results to the frontend, including:
- insight (string): The paraphrased finding
- source (string): Citation ID (e.g., "source-001") or source reference
- quotes (list of strings): Supporting quotes from the source

The system SHALL NOT include confidence in the frontend serialization.