refactor: cave and river (#22)

* refactor(chunk): add ChunkInfo and switch to shared_mutex for chunk access

Introduced ChunkInfo struct to expose chunk metadata (position, seed, biome). Replaced std::mutex with std::shared_mutex for chunk map to allow concurrent read access. Added World::get_chunk_info() method. Temporarily disabled cave/river cleanup and debug biome reporting.

* refactor(cave,river,chunk): use ChunkPos as key for paths and track cave existence

* refactor(gameplay): use deterministic origin-based cave and river generation

Compute cave and river paths on-the-fly per chunk from a deterministic origin instead of storing them globally. Remove concurrent hash map storage, shared mutexes, and related cleanup methods. This simplifies concurrency and reduces memory overhead.

* refactor(world): remove unused chunk generation progress tracking

include/Cubed/gameplay/chunk_generator.hpp

@@ -4,7 +4,6 @@
 #include "Cubed/gameplay/biome.hpp"
 #include "Cubed/gameplay/block.hpp"
 #include "Cubed/gameplay/builders/biome_builder.hpp"
-#include "Cubed/gameplay/path_point.hpp"
 #include "Cubed/tools/cubed_random.hpp"
 
 #include <atomic>
@@ -62,9 +61,6 @@ private:
     unsigned m_chunk_seed = 0;
 
     void make_biome_builder();
-    void
-    carve_worm(const std::vector<PathPoint>& points, const ChunkPos& chunk_pos,
-               std::function<void(int /*x*/, int /*y*/, int /*z*/)> on_hit);
 };
 
 } // namespace Cubed