fix: reinit wall.drawable_count in StartKyoku replay handler

[smly]

Fixes #198.

• Resets wall.drawable_count to wall.tiles.len() - 14 inside the StartKyoku arm of apply_mjai_event so it matches the invariant established by _initialize_round. Applies the same fix to the 3P replay handler.
• Adds regression coverage.

[Copilot]

Pull request overview

Fixes a replay-only state initialization bug where StartKyoku rebuilt the wall tiles but left wall.drawable_count stale, which could incorrectly suppress Riichi (and other actions gated by remaining draws) in observe_event replays.

Changes:

• Reinitialize wall.drawable_count in apply_mjai_event(StartKyoku) to wall.tiles.len() - 14 for both 4P and 3P replay handlers to match _initialize_round.
• Add regression tests asserting drawable_count resets correctly and that Riichi appears in legal actions in replay scenarios for both variants.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
riichienv-core/src/state/event_handler.rs	Reset wall.drawable_count during 4P StartKyoku replay initialization to restore legality invariants.
riichienv-core/src/state_3p/event_handler.rs	Same drawable_count reset for 3P StartKyoku replay initialization.
riichienv-core/src/tests.rs	Add regression coverage for drawable_count reset and Riichi availability in replay paths (4P/3P).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.