Delete tavern gamble tests that hand-roll the payout instead of calling gamble()

[dmccoystephenson]

Summary

• test_gamble_win_shows_correct_amount and test_gamble_loss (tests/location/test_tavern.py) hand-executed the win/loss arithmetic instead of calling Tavern.gamble() — both pass even with gamble() deleted entirely (confirmed by temporarily stubbing it out and rerunning).
• The win test additionally encoded the wrong payout: even money (money == 150 on a $50 bet) instead of the real DICE_WIN_MULTIPLIER (5x) payout of $350 — a hardcoded assertion that could never catch a regression to that old behavior.
• Deleted both. test_gamble_win_pays_multiple_of_bet and test_gamble_loss_via_real_loop already drive the real gamble() loop and cover the same branches correctly; folded the win test's "shows the actual amount, not $0" prompt-text check into test_gamble_win_pays_multiple_of_bet, deriving the expected figure from DICE_WIN_MULTIPLIER instead of hardcoding it.

Test plan

• SDL_VIDEODRIVER=dummy SDL_AUDIODRIVER=dummy python3 -m pytest tests/location/test_tavern.py — 30 passed
• SDL_VIDEODRIVER=dummy SDL_AUDIODRIVER=dummy python3 -m pytest — 727 passed
• Regression check: temporarily changed winAmount = self.currentBet * DICE_WIN_MULTIPLIER to winAmount = self.currentBet (reintroducing the even-money bug) — test_gamble_win_pays_multiple_of_bet failed as expected, confirming the surviving test actually exercises the payout math. Reverted afterward.

Flagging for review: this PR deletes 78 lines from a single file, over this repo's 50-line do-not-auto-merge threshold, so per the dev-loop's merge policy it needs an explicit human decision rather than an autonomous merge even though it's test-only and CI is green.

Closes #154

---

drafted by Claude on behalf of Daniel Stephenson

[dmccoystephenson]

Self-review rubric:

• Scope: PASS — only tests/location/test_tavern.py touched, matches test: two tavern gamble tests re-implement the payout instead of calling gamble(), and encode the wrong multiplier #154 exactly.
• Tests-new: N/A — this is a deletion/consolidation of existing test coverage, not new production code.
• Tests-fix (empirical): PASS — reintroduced the even-money bug in src/location/tavern.py (winAmount = self.currentBet instead of * DICE_WIN_MULTIPLIER) and reran; test_gamble_win_pays_multiple_of_bet FAILED as expected. Reverted the source change (git checkout -- src/location/tavern.py) and confirmed 727/727 pass again.
• Sibling structure: N/A — no new files.
• Sibling renames: N/A.
• Docs: PASS — no doc references the two deleted test names.
• Issue resolution: PASS — both hand-rolled tests named in test: two tavern gamble tests re-implement the payout instead of calling gamble(), and encode the wrong multiplier #154 are deleted; the worthwhile prompt-text assertion was preserved by folding it into test_gamble_win_pays_multiple_of_bet, derived from DICE_WIN_MULTIPLIER rather than hardcoded, per the issue's suggested fix.
• CI: PASS — test check green on 34e46a1 (https://github.com/Stephenson-Software/FishE/actions/runs/30771473878).
• Schema-sync: N/A.
• Money-format: N/A.
• Deterministic-tests: PASS — new/changed assertions use patch()-scoped random.randint, no unseeded random.
• Headless-pygame: N/A.
• camelCase: N/A — no new identifiers.

Summary: all rubric items pass, CI green, fix empirically verified by reverting and re-testing. Do-not-auto-merge match: this diff deletes 78 lines from a single file (over the repo's 50-line threshold), so per Phase 8 this needs an explicit human merge decision rather than an autonomous merge.

---

drafted by Claude on behalf of Daniel Stephenson