Add a Markdown reporter for PR comments and CI summaries

[AnshBajpai05]

Third slice from #80, built on the comparison/reporter split (#90).

What it adds

--markdown report.md — a GitHub-flavored report you can drop straight into a PR comment or $GITHUB_STEP_SUMMARY:

oracletrace app.py --markdown report.md
oracletrace app.py --compare baseline.json --markdown report.md
oracletrace run -- pytest tests/ --markdown report.md

Contents:

• summary line (total time, function count, timestamp)
• when --compare is given: a comparison table with 🔴/🟢 per changed function, new/removed function lists, and a regression banner (🚨 count / ✅ none)
• top functions by total time

Example (a real run against a shrunk baseline):

### Comparison vs. baseline

| Function | Baseline | Current | Change |
|---|---:|---:|---:|
| `s.py:slow` | 0.0012s | 0.0039s | 🔴 +238.4% |

> 🚨 **1 regression(s)** above 5.0% threshold

How it stays clean

The reporter is a pure consumer of ComparisonData from #90 — it re-implements none of the comparison logic. That's exactly what the split was for: render_markdown(data, comparison, top) reads statuses and renders; a future JUnit or other reporter is the same shape.

Function names are escaped (| and backtick) so a name can't break out of a table cell or inline code.

Wiring

--markdown is added to the base parser and to baseline save, mirroring --html. To let the Markdown report include the comparison, the compare step was split so the result is available at export time: _load_comparison() (load baseline + compare) → _export_results(data, args, comparison) → _report_comparison() (print + regression gate). No behavior change for existing flags — the terminal comparison, exit-2 gate, and every other export are unchanged; comparison errors still return exit 1 with the same messages.

Tests

• test_markdown_reporter.py — header/summary, top table sorting + --top limit, pipe-escaping, comparison table with icons, regression/no-regression banners, new/removed lists, file writing, self-contained (only the footer link).
• test_cli.py — --markdown writes a file end-to-end, and --compare --markdown includes the comparison section.

Full suite green locally (the one failure on my machine is the pre-existing test_symlinked_user_code, which needs Windows symlink privileges and passes on CI).

🤖 Generated with Claude Code