Expand unit tests for milestone/achievement helpers

[dmccoystephenson]

Summary

Stage B (unit-test expansion) cycle targeting src/achievements/achievements.py. Before this PR the module (238 lines) was covered by 4 tests in tests/achievements/test_achievements.py. arcadeIdForName had no direct test, and the integrity of the MILESTONES data table was not checked anywhere.

15 characterization tests are added. Production code is not changed.

• Data-integrity guards on MILESTONES: every stat is a real Stats attribute. This matters because isEarned falls back to 0 through getattr, so a typo in stat would make a milestone silently impossible to earn. The guards also check that milestone names are unique and do not collide with GOAL_MILESTONE_NAME (these names are what stats.earnedMilestones persists), and that every threshold is positive and every description is non-empty.
• isEarned: a missing stat is not earned.
• getMilestoneStatuses: the order matches MILESTONES, earned flags are correct, and nothing is recorded on stats.earnedMilestones.
• getNewlyEarned: several milestones earned at once come back in MILESTONES order. Milestones already recorded by a loaded save are skipped. A fresh game returns nothing.
• arcadeIdForName: maps the goal, maps a milestone that has an arcade id, and returns None for a milestone without an arcade id and for an unknown name.
• arcadeDeclarations: milestones come first, then the goal, then retirement, with exact goal/retired entries. Title, description and hidden flags come from each milestone. This complements the id-validity checks already in tests/achievements/test_arcade.py.

No tracking issue: the issue backlog was empty at triage, so this cycle was given to test expansion. Every new test passed against the unchanged code, and no bug was found.

Test plan

• python3 -m pytest tests/achievements/test_achievements.py: 19 passed (4 existing + 15 new)
• Full suite locally (dummy SDL drivers): 979 passed, 1 failed. The failure is tests/test_saveFileManager.py::test_migrate_old_save_files_skips_a_slot_holding_any_file. It is caused by the sandbox's locally installed editable tak 0.1.1 instead of the commit pinned in requirements.txt, and is unrelated to this test-only diff. CI installs the pinned tak and is the authoritative anchor.
• CI Run Unit Tests green on the PR head

This PR description was drafted during a Gardener session (https://github.com/Stephenson-Software/gardener).

🤖 Generated with Claude Code

---

drafted by Claude on behalf of Daniel Stephenson

[dmccoystephenson]

Self-review rubric (Stage B, test-only diff):

• Scope: PASS. gh pr diff 222 touches only tests/achievements/test_achievements.py (+219 lines, all new test functions). No production code, formatting churn or renames.
• Tests-new: PASS (not applicable as a gate). No new public functions were added. All 15 new tests target existing functions in src/achievements/achievements.py; arcadeIdForName had no direct test before this PR.
• Tests-fix: not applicable. No bug fix is included, and every new test passes against the unchanged production code (characterization tests).
• Sibling structure: PASS. The new tests follow the file's existing createStats() factory and the # prepare / # call / # check layout, with test_<behavior> names.
• Sibling renames: not applicable. Nothing was renamed.
• Docs: PASS. The Phase 7 table was checked: no README feature, schema, PLANNING goal or version is affected by a test-only change.
• Issue resolution: not applicable. There is no Closes reference; the issue backlog was empty at triage.
• CI: PASS. Run Unit Tests on the PR head ran 980 tests, all passing (run 37902542304). The "Upload coverage reports" step logs a Codecov tokenless-upload error, but the job still succeeds; this matches the run on main and carries no signal for this diff.
• Schema-sync: not applicable. No persisted fields were touched.
• Money-format: not applicable. No displays were touched.
• Deterministic-tests: PASS. The new tests use no randomness; they set Stats attributes directly.
• Headless-pygame: not applicable. No pygame code path is involved.
• camelCase: PASS. Locals such as expectedMilestoneIds and byId follow camelCase, as in the existing tests.

Judgment call flagged for the reviewer:

• tests/achievements/test_achievements.py — test_every_milestone_has_a_positive_threshold_and_a_description makes it an invariant that no milestone has a threshold of 0. Today's data satisfies this. A future deliberate "free on day one" milestone would need to update this test.

Out-of-diff observation: locally, tests/test_saveFileManager.py::test_migrate_old_save_files_skips_a_slot_holding_any_file fails against the sandbox's editable tak 0.1.1. CI, which installs the tak commit pinned in requirements.txt, passes it. This is environmental and has no bearing on this PR.

Summary: a test-only expansion of the achievements helpers. CI is green with 980 tests, and the PR is ready to merge.

This review was drafted during a Gardener session (https://github.com/Stephenson-Software/gardener).

---

drafted by Claude on behalf of Daniel Stephenson