Fix Simulator for west-round terminal states

[tsubakisakura]

Summary

• Simulator.get_next_state() の終局判定を zero_indexed_kyoku ベースに修正
• W2 など東場以外での最終盤面を正しく終局扱いできるようにした
• start_game/end_game のみ（start_kyoku なし）の continue 結果を正常終了として扱うガードを追加

Why

• seed=(10030,25030) で run_0 終了後に不要な run_1 が走り、RuntimeError: can't start game が発生していた
• 原因は Python 側の終局判定条件が不十分で、終局済みでも continue し得ること

再現コード

#!/usr/bin/env python3
import mjai

submissions = [
    "examples/rulebase.zip",
    "examples/rulebase.zip",
    "examples/rulebase.zip",
    "examples/rulebase.zip",
]

seed = (10030, 25030)
mjai.Simulator(submissions, logs_dir="./logs", seed=seed).run()

[Copilot]

Pull request overview

This PR fixes Simulator.get_next_state() terminal-state detection in continue mode by basing the “final hand and later” check on a zero-indexed kyoku value, ensuring West-round states (e.g., W2) are handled as terminal when appropriate, and adds a guard to treat start_game/end_game-only logs as a clean finish.

Changes:

• Add a guard to return None (clean finish) when a continue-mode log contains only start_game/end_game events.
• Convert (bakaze, kyoku) to a unified zero_indexed_kyoku earlier and use it for terminal (“オーラス以降”) detection.
• Remove redundant late conversion by reusing zero_indexed_kyoku for subsequent next-state computation.

---

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

[Copilot]

event_types.issubset({"start_game", "end_game"}) is also True for an incomplete log that contains only start_game (or even an empty events list), which would be treated as a clean finish and could mask real failures. Consider requiring that end_game is actually present (e.g., check that event_types == {"start_game","end_game"} or at least that both types exist) before returning None.

Suggested change

	if event_types.issubset({"start_game", "end_game"}):
	if event_types.issubset({"start_game", "end_game"}) and "end_game" in event_types:

[smly]

指摘ありがとうございます！get_next_state() が呼ばれる条件を精査します。

[Copilot]

This changes the terminal-state logic to use zero_indexed_kyoku >= 7 (covering West-round hands too), but there doesn’t appear to be a regression test that exercises a West-round continuation (e.g., W1/W2) and asserts get_next_state() returns None when it should. Adding a focused unit/integration test in tests/test_game_simulator.py for a West-round terminal scenario (and the new start_game/end_game-only case) would help prevent future regressions.

[smly]

@tsubakisakura これは恥ずかしい見落とし... 修正ありがとうございます！Copilot のコメントしている regression test は不要です。