Added the stremlit UI for tailoring the AI resume

[Devesh-Maheshwari]

Summary

Wraps the existing V2 pipeline in a Streamlit web app so the tool is usable by
anyone, not just people who can run a Colab notebook. The underlying pipeline
functions are unchanged — they were refactored out of the notebook into an
importable module (pipeline.py) and driven by a UI.

Closes #5

What changed

• pipeline.py — the notebook's steps as importable functions
  (fetch_github_repos, chunk_source_material, extract_jd_requirements,
  retrieve_relevant_experience, generate_tailored_resume,
  evaluate_rag_coverage), plus the fabrication check, injection screen, and
  diff. Logic preserved; only parameterized to drop Colab-only bits
  (userdata, files.upload, input()). Added a run_pipeline() generator
  that yields progress events, retry/503 handling, and user-facing errors.
• app.py — Streamlit UI.
• requirements.txt, .streamlit/config.toml, tests/test_pipeline.py (20 tests), .gitignore.

UX from the issue, addressed

• Drag-and-drop resume upload (.md/.txt) with a paste fallback
• Rendered markdown output (not raw print), with a [SOURCE]-tag toggle and Markdown download
• Coverage dashboard — metrics, progress bar, per-requirement evidence, honest gaps
• Live progress states via st.status (each pipeline step streams as it runs)
• Clean error states — the 503 case and GitHub-fetch failure surface as friendly
  messages instead of stack traces (GitHub failure is non-fatal; the run continues)

Verification

• 20/20 unit tests pass (python -m pytest tests/)
• Full end-to-end run against Gemini: tailored resume generated with source tags,
  100% requirement coverage on the sample, fabrication check ran
• Streamlit app boots healthy (health endpoint returns ok)

Run locally

cd ai_resume_builder
pip install -r requirements.txt
# provide a Gemini key via .env (GOOGLE_API_KEY=...), Streamlit secrets, or env var
streamlit run app.py

[Devesh-Maheshwari]

@camunity Can you please review it