feat: Add K-Nearest Neighbor (<->) operator for blazing fast spatial searches

[seuros]

Summary

This PR adds the PostGIS K-Nearest Neighbor (<->) operator to the gem, enabling lightning-fast spatial searches that utilize spatial indexes.

Why This Matters

Traditional "find nearest" queries using ST_Distance + ORDER BY scan the entire table and calculate distances for every row. The <-> operator uses spatial indexes to return results orders of magnitude faster.

Implementation

New Arel Methods

• distance_operator(geometry) - The primary method for KNN searches
• :<-> alias - For those who prefer symbolic operators

Usage Examples

# Find 10 nearest locations (uses spatial index\!)
Location
  .order(Location.arel_table[:position].distance_operator(my_position))
  .limit(10)

# Find nearest restaurants with actual distances
Restaurant
  .select("*, ST_Distance(location, ST_GeomFromText('#{origin.as_text}', 4326)) as distance")
  .order(Restaurant.arel_table[:location].distance_operator(origin))
  .limit(5)

# Emergency: Find nearest hospitals
Hospital
  .where(emergency_room: true)
  .order(Hospital.arel_table[:coordinates].distance_operator(accident_location))
  .first

Performance Impact

The <-> operator can be 100-1000x faster than traditional distance queries for large datasets because it uses the spatial index directly rather than calculating distances for every row.

Documentation

Updated both the technical documentation and the space-themed guides with examples of tactical uses for the KNN operator.

---

That's it folks! This completes the extraction of custom spatial code I had in my projects. Any additional features can be added via PRs with tests.