Feat/generate user facing nlp explanations

[AAdewunmi]

Summary

Adds user-facing NLP explanations for StudyBuddy insights.

Changes

• Added deterministic explanation generation in apps/insights/nlp/explanations.py.
• Added direct explanation tests in apps/insights/tests/test_explanations.py.
• Updated docs/ai-nlp-contract.md to document explanation behaviour as current test coverage.
• Added Sprint 3 Day 3 console verification runbook.

Behaviour

• Explains the confidence label used for an insight.
• Shows how many meaningful terms were analysed.
• States that keywords are ranked by deterministic term frequency.
• States that summaries use original note sentences rather than generated claims.
• Lists detected keywords.
• Handles missing keywords with no strong repeated terms.
• Handles low-information input by asking for more detailed notes instead of overstating insight quality.

Why

Explanations keep the NLP feature transparent and honest. They help users understand that insights come from deterministic text processing, not model intelligence or unsupported generated claims.

Validation

• pytest apps/insights/tests/test_explanations.py -q
• pytest apps/insights/tests/test_summarisation.py apps/insights/tests/test_confidence.py apps/insights/tests/test_explanations.py -q
• pytest apps/insights -q
• pytest -q
• ruff check apps/insights/tests/test_explanations.py
• black --check apps/insights/tests/test_explanations.py
• ./docs/sprint-runbook/sprint-3/sprint-3-day-3.sh

Result

All targeted NLP tests, insights tests, full regression tests, lint checks, format checks, and the Sprint 3 Day 3 runbook pass.

Notes

This remains deterministic NLP. The explanation text does not imply LLM use, semantic understanding, probability, or factual correctness.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!