feat(searching): add k closest points to origin (heap and quickselect)

[0xDevNinja]

Summary

• Implement two algorithms for finding the k closest 2D points to the origin in src/searching/k_closest_points.rs:
  • k_closest_heap — bounded max-heap of size k keyed by squared distance, O(n log k) time, O(k) extra space.
  • k_closest_quickselect — Hoare-style in-place partition with a deterministic xorshift64* pivot seeded by points.len(); O(n) average / O(n^2) worst time, O(n) extra space.
• Both routines compare squared distances (no sqrt), share NaN-tolerant ordering, and document that returned-vector order is unspecified with ties broken by encounter order.
• Edge cases handled: k == 0 returns empty, k >= n returns all, empty input returns empty.

Closes #381

Test plan

• cargo fmt
• cargo build
• cargo clippy --all-targets -- -D warnings
• cargo test (2668 unit + 51 doctests pass; 10 new unit tests cover k=0, empty input, k > n, k == n, simple known example, single point, cross-impl agreement on small inputs, count invariant across all k, brute-force distance multiset check, and tied-distance behavior)