feat: add heavy-load CPU/memory profiling mode

[pandego]

Summary

This PR adds built-in runtime profiling support for heavy-load runs so users can measure end-to-end latency and memory usage directly from the CLI.

What changed

• Added pageindex/profiling.py with:
  • profile_run(...) context manager
  • write_profile_report(...) JSON report writer
• Extended run_pageindex.py with:
  • --enable-profile
  • --profile-output (default: ./results/profile_report.json)
• Updated README with a heavy-load profiling walkthrough and report interpretation guidance.
• Added tests/test_profiling.py coverage for report generation/writing logic.

Why

Issue #141 asks for practical CPU/memory profiling documentation under heavy load. This PR provides both:

• executable profiling support in the main entrypoint,
• and docs showing how to use it for repeatable performance analysis.

Validation

• python3 -m py_compile run_pageindex.py pageindex/profiling.py tests/test_profiling.py

Note: full runtime tests were not executed in this environment because Python package dependencies/tooling (pip/pytest) are unavailable on this runner.