docs: add notes on JPS, landmarks

docs/primitives/astar.md

@@ -2,9 +2,17 @@
 
 The [venerable A\* algorithm](https://www.redblobgames.com/pathfinding/a-star/introduction.html) provides the baseline for pathfinding in Screeps.
 
-Red Blob Games has some [WIP articles on heuristics](https://www.redblobgames.com/blog/2024-05-05-wip-heuristics/).
+Within a single room, the Clockwork implementation of the A\* algorithm alone is competitive with PathFinder. However, across longer distances (greater than a single room), PathFinder's JPS optimizations make a significant difference.
+
+We'll experiment with a few different options.
+
+## Optimizations
+
+- [Jump-Point Search](https://zerowidth.com/2013/a-visual-explanation-of-jump-point-search/) is also used under the hood by PathFinder
+- [Pre-computed landmarks](https://www.redblobgames.com/blog/2024-05-05-wip-heuristics/) for more accurate A\* heuristics
+
+## Other interesting variants
 
 - [Cooperative Pathfinding](https://theory.stanford.edu/~amitp/GameProgramming/MovingObstacles.html#predicting-obstacle-movement) (building a table of the paths of other units)
-- JPS (generating paths, ignoring terrain costs)
 - [Incremental heuristic search](https://en.wikipedia.org/wiki/Incremental_heuristic_search) (tracking moving targets)
 - [Parallel Breadth-First Search](https://arxiv.org/abs/2210.16351)

docs/wasm/index.md

@@ -3,3 +3,4 @@
 We are using wasm-bindgen to create the WASM and a JS shim (`packages/wasm`). There are a few things we tweak to improve performance (such as passing in the `__packedPos` instead of the whole RoomPosition), so this is further wrapped in a Typescript library (`packages/screeps-clockwork`) which wraps and re-exports the WASM functionality.
 
 - Copying data across the [WASM boundary](./boundary.md) adds up, so we must be thoughtful about when and how we copy data back and forth.
+- Javascript can't garbage-collect our Rust objects, so we need to [clean them up explicitly](./cleanup.md).