Feat/add extractive summary generator

[AAdewunmi]

Summary

Adds a deterministic extractive summary generator for StudyBuddy insights.

Changes

• Added apps/insights/nlp/summarisation.py.
• Implemented summarise_text() for source-grounded summaries.
• Added LOW_INFORMATION_SUMMARY fallback text.
• Added tests in apps/insights/tests/test_summarisation.py.

Behaviour

• Splits note text into source sentences.
• Scores sentences using meaningful tokens and extracted keyword frequency.
• Selects high-signal source sentences.
• Returns selected sentences in original source order.
• Returns a low-information message for empty, sparse, or unusable input.
• Does not generate new claims or rewrite user notes.

Why

This keeps summaries deterministic, explainable, and grounded in the user’s own study notes, matching the AI/NLP contract for the Sprint 3 MVP.

Validation

• pytest apps/insights/tests/test_summarisation.py apps/insights/tests/test_nlp_keyword_extraction.py apps/insights/tests/test_nlp_text_processing.py -q
• TEST_DATABASE_URL=postgres://studybuddy:studybuddy@localhost:5432/studybuddy_test pytest apps/insights -q
• TEST_DATABASE_URL=postgres://studybuddy:studybuddy@localhost:5432/studybuddy_test pytest -q --reuse-db
• black . --check
• ruff check .

Result

Extractive summary generation is implemented and covered by focused tests. The PostgreSQL-backed insights and full regression suites pass.

Notes

This adds summary generation only. Confidence scoring, explanation generation, persistence reuse, and views remain separate follow-up work.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!