Establish comprehensive testing foundation for future UI evolution:

**Test Framework**:
- Playwright for E2E browser testing
- Configured for Chromium (extensible to Firefox/Safari)
- HTML reporting with traces and screenshots
- CI-ready with automatic retries

**Test Coverage** (`tests/playwright/cesium-queries.spec.js`):
- ✅ Page loading and geocode search box
- ✅ Camera movement on geocode search
- ✅ HTML table structure (5 columns: Thumbnail | Sample | Description | Site | Location)
- ✅ Clickable sample PID links to OpenContext
- ✅ "View site" links
- ✅ Thumbnail images and "No image" placeholders
- ✅ Result count displays
- ✅ Empty state friendly messages
- ✅ Scrollable tables with sticky headers
- ✅ Zebra-striped rows
- ✅ Visual consistency across all three tables

**Test Data**:
- PKAP location with samples: `geoloc_04d6e816218b1a8798fa90b3d1d43bf4c043a57f`
- Larnaka site marker (empty state): `geoloc_7a05216d388682536f3e2abd8bd2ee3fb286e461`

**Infrastructure Files**:
- `playwright.config.js`: Test configuration with extended timeouts
- `package.json`: NPM scripts (test, test:headed, test:ui, test:debug)
- `tests/README.md`: Comprehensive testing guide
- `tests/playwright/cesium-queries.spec.js`: Full test suite

**NPM Scripts**:
```bash
npm test              # Run all tests
npm run test:headed   # Run with browser visible
npm run test:ui       # Interactive UI mode
npm run test:debug    # Debug mode with inspector
npm run test:report   # View HTML report
```

**Gitignore Updates**:
- node_modules/
- test-results/
- playwright-report/
- package-lock.json

**Future-Ready**:
This infrastructure provides a solid foundation for:
- Adding new UI feature tests
- Visual regression testing
- Accessibility testing
- Performance monitoring
- Cross-browser testing
- Mobile responsive tests

Tests validate the enhanced query UI (Path 1, Path 2, Eric's query) to ensure consistent, high-quality user experience as the UI evolves.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

…nhancement

**Performance Improvement**: 7+ seconds → ~2 seconds (71% faster!)

## Problem
The expensive CTE query with JOIN + GROUP BY took 7+ seconds to classify
each geocode as sample_location/site_location/both before rendering any dots.
This made the page feel frozen and unresponsive.

## Solution: Progressive Enhancement
1. **Fast initial load** (~2s): Simple SELECT DISTINCT query, all dots blue
2. **Optional refinement** (~7s): Button to classify and color-code by type
3. **Chunked rendering**: 500 points per batch with progress indicator
4. **Performance telemetry**: Console logs show timing for all operations

## Technical Changes

### Simplified Initial Query (lines 137-163)
**Before** (expensive):
```sql
WITH geo_classification AS (
    SELECT geo.pid, geo.latitude, geo.longitude,
           MAX(CASE WHEN e.p = 'sample_location' THEN 1 ELSE 0 END) as is_sample_location,
           MAX(CASE WHEN e.p = 'site_location' THEN 1 ELSE 0 END) as is_site_location
    FROM nodes geo
    JOIN nodes e ON (geo.row_id = e.o[1])
    WHERE geo.otype = 'GeospatialCoordLocation'
    GROUP BY geo.pid, geo.latitude, geo.longitude
)
SELECT pid, latitude, longitude, CASE ... END as location_type
FROM geo_classification
```

**After** (fast):
```sql
SELECT DISTINCT pid, latitude, longitude
FROM nodes
WHERE otype = 'GeospatialCoordLocation'
```

### Optional Classification Button (lines 50-56, 769-845)
- User can click button to run classification query
- Updates existing point colors/sizes in-place
- Same color scheme as before:
  - Blue (small): sample_location_only - field collection points
  - Purple (large): site_location_only - administrative markers
  - Orange (medium): both - dual-purpose locations

### Chunked Rendering (lines 169-218)
- Render 500 points per batch
- Yield to browser between chunks (keeps UI responsive)
- Dynamic progress: "Rendering geocodes... 500/198,433 (0%)"

### Performance Telemetry (lines 376-384, 438-446, 500-508)
- Console logs for all queries: locations, Path 1, Path 2, Eric's query
- Example output:
  ```
  Query executed in 1847ms - retrieved 198433 locations
  Rendering completed in 423ms
  Total time (query + render): 2270ms
  ```

## User Experience Improvements

**Before**:
- Page frozen for 7+ seconds with static "Loading..." text
- No feedback on progress
- User uncertain if page crashed

**After**:
- Interactive in ~2 seconds with all dots visible
- Progress indicator shows "Rendering geocodes... X/Y (Z%)"
- Optional color-coding button if user wants classification
- Console telemetry for debugging and optimization planning

## Files Changed
- `tutorials/parquet_cesium.qmd` (+86 lines): Optimized queries + telemetry
- `OPTIMIZATION_SUMMARY.md` (new): Performance analysis and results
- `LAZY_LOADING_IMPLEMENTATION.md` (new): Technical implementation details
- `PERFORMANCE_OPTIMIZATION_PLAN.md` (new): Future optimization roadmap

## Testing
Test at http://localhost:XXXX/tutorials/parquet_cesium.html
- Expect: Page loads in ~2 seconds with all blue dots
- Click "Color-code by type" button → dots recolor after ~7 seconds
- Console shows timing for all queries

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

The button wasn't appearing on the page because echo: false was hiding both
the code and preventing proper Observable rendering. Using echo: fenced shows
the code in a collapsible section while ensuring the button UI renders.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>