Decode reward keys and use mapping files

Add decoding for Monopoly reward keys and switch reward metadata to be looked up by decoded id. Implement decode_reward_key() and infer_reward_type(); replace the previous raw-key KNOWN_REWARDS approach with REWARDS_BY_ID built from mappings/*.json. Add mappings/items.json and expand mappings/arcs.json. Update decoder to produce reward_id, use REWARDS_BY_ID, and change guess_quantity() to accept reward_id. Refactor pool_mappings to reuse the new mapping loader, and update tests and docs to validate mapping files and document the reward key encoding/behavior.

docs/limitations.md

@@ -12,6 +12,6 @@ Known limitations:
 - Live capture currently uses Windows raw sockets and requires administrator permission.
 - The file adapter reads mitmproxy `.flows` captures for research and testing.
 - The more stable and reliable Npcap/libpcap capture is not implemented yet.
-- Reward mappings are research mappings and should be expanded as more samples are decoded.
+- Reward keys decode to their reward id string, so unknown rewards still export a usable `reward_id`; display names/ranks come from the mapping JSON files and should be expanded as new rewards appear.
 
 Privacy guardrail: raw packet data must not be included in sanitized exports.