Add “First-Glance” Statistics for Search Results

We should enrich the search results page with lightweight exploratory statistics to help users understand patterns in the matched corpus.

**Proposed additions:**

- Keyword distribution (including partial-word matches and regex cases).
- Frequent previous/next tokens around each hit (simple ±1 context).
- Timeline distribution of matches (histogram/bar plot or compact sparkline).
- Optional: top documents, speaker/source breakdown, average snippet length.

**Motivation:**
These quick-look stats will make the interface more informative and help identify patterns before diving into detailed results. Useful for both exploratory research and debugging queries.

**Implementation notes:**

- Should run fast on the indexed results (no deep re-scan).
- Keep computation modular so each statistic can be toggled on/off.
- Consider exporting the stats as JSON for future visualization.

Start at the search results page and the indexed-results path described in the issue. Define the statistic set and toggles before implementing; done means fast, modular keyword, context, and timeline statistics are available, with optional JSON export if included.

Written by the indexing model from the issue text.