fix: recover progress from a last-known-good state backup

[abhiksark]

Motivation

Progress is valuable to a learner. The current recovery path renames a corrupt primary state file onto state.json.bak, which can overwrite the only valid recovery point and then starts fresh instead of attempting recovery.

Current evidence

• pythonlings/core/state.py::load() renames an unreadable state.json to state.json.bak and returns fresh state.
• That behavior can overwrite an existing valid backup.
• docs-site/faq.md describes where progress lives and needs to match the recovery contract.

Exact scope

Maintain a valid last-known-good .pythonlings/state.json.bak, recover it when the primary state is corrupt, and quarantine bad primary data without overwriting the backup. Align FAQ wording and tests with the final behavior.

Acceptance checklist

• A successful state save preserves a valid last-known-good backup.
• If the primary state is corrupt and the backup is valid, loading recovers the backup state and tells the user what happened.
• Corrupt primary data is quarantined with a distinct, non-destructive name; it never replaces a valid backup.
• If both files are unusable or no backup exists, recovery starts fresh with a friendly warning and preserves available forensic data.
• State save/recovery remains atomic enough that normal progress is never silently lost.
• Unit tests cover valid backup creation, successful recovery, corrupt backup, and absent backup; the FAQ describes the same behavior.

Explicit exclusions

Do not add cloud sync, change the state format version, delete corrupt data automatically, or alter unrelated workspace migration behavior.

Likely files

• pythonlings/core/state.py
• tests/unit/test_state.py
• docs-site/faq.md

Verification

python -m pytest tests/unit/test_state.py -q
python -m pytest -q
mkdocs build --strict

Manually corrupt a primary state after creating valid progress and confirm the recovered progress is visible without overwriting state.json.bak.

Sprint coordination

See the sprint Discussion. This issue is unassigned and is not a reservation.

[coderabbitai]

🔗 Related PRs

#31 - feat: overall progress, completion celebration, first-launch welcome (#14, #15, #16) [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!