fix(backend): update core files for correct SMILE methodology

[jv-singh]

Summary

• The previous codebase hardcoded a hallucinated 5-phase acronym (SENSE → MODEL → INTERVENE → LEARN →
  EVOLVE) that does not exist in the LPI data source. The real SMILE is Sustainable Methodology for Impact
  Lifecycle Enablement — 6 sequential phases sourced verbatim from data/smile-framework.json.

• This PR replaces the 5-phase enum with the correct 6-phase enum, recomputes the scoring formula, rewrites
  the SMILE and scoring tests for the new phases, and fixes three documentation files that still contained
  literal "sense" strings — anyone copying the QA matrix or running the proxy-data script would have hit 422
  validation errors otherwise.

• Phases (correct, in order):

1. reality-emulation — establish the reality canvas
2. concurrent-engineering — define the MVT, validate virtually
3. collective-intelligence — sensors, ontologies, KPIs
4. contextual-intelligence — real-time decisions, connected twin
5. continuous-intelligence — AI prognostics, simulation
6. perpetual-wisdom — share impact, circular strategies

The transition rule (forward one step allowed, backward any, skip forbidden, same-phase forbidden) is
unchanged — only the phases it operates on were wrong.

Score range: [0.80, 6.70] → [0.80, 7.00] (PERPETUAL_WISDOM = 6, the new max weight).

Closes the test failures in tests/test_smile.py and tests/test_scoring.py caused by phase-name mismatches.

Type

• Fix (bug — wrong phase enum, wrong weights, wrong test expected values, stale doc literals)
• Refactor (replaced 5-phase enum and PHASE_WEIGHTS with the correct 6-phase versions; added
  get_phase_key_question())
• Docs / reports (3 doc files with stale "sense" literals corrected)

Checklist

• make lint exits 0 (run locally before push)
• make test exits 0 (run locally before push)
• Tests added or updated for changed behaviour (tests/test_smile.py, tests/test_scoring.py,
  tests/test_smoke.py, tests/test_goal_crud.py, tests/conftest.py)
• Daily report updated in reports/ (see note below — needs to be filled in before merge)
• No .env secrets committed
• PR targets staging (never main directly) → note: current branch is fix/backend-smile-updates, target
  staging

Tests passing

• tests/test_smoke.py — test_imports, test_health, test_smile_phase_order (now asserts 6 phases)
• tests/test_smile.py — TestPhaseOrder (5 tests), TestPhaseTransitions (15 tests), TestPhaseDescriptions
  (6 tests)
• tests/test_scoring.py — TestPhaseWeightsConstant (13 tests), TestScoreGoalKnownValues (12 tests),
  TestScoreGoalWithUrgency (4 tests), TestScoreGoalProperties (5 tests), TestSortGoalsByScore (9 tests),
  TestScoreExplanation (5 tests) — 48 tests total, up from 46
• tests/test_goal_crud.py — TestCreateGoal (4), TestReadGoal (4), TestUpdateGoal (4), TestDeleteGoal (1) —
  all phase literals now correct
• tests/test_recommendations.py — unaffected (no phase string assertions, only docstring mentions)

Daily report reference

reports/2026-06-06.md (to be created — covers the SMILE 6-phase correction and the doc-only follow-up)

Files changed

LPI_SMILE_Written_Report.pdf

[jv-singh]