feat(calibration): render backtest score/comparison reports as Markdown

[kai392]

Summary

• Add packages/loopover-engine/src/calibration/backtest-report.ts with renderBacktestScoreReport and renderBacktestComparison: the deterministic, pure Markdown "receipt" renderers for BacktestScoreReport (calibration: pure confusion-matrix scorer for a candidate rule classifier against the backtest corpus #8085) and BacktestComparison (calibration: Pareto-floor comparator between two BacktestScoreReports #8086) — byte-identical output for byte-identical input, no IO, no wall-clock reads.
• renderBacktestScoreReport renders the rule ID, caseCount, all four confusion-matrix counts, and precision/recall as a Markdown table, with null ratios rendered as the literal N/A — never 0, the word null, or an empty cell (the null-is-not-zero discipline from calibration: pure confusion-matrix scorer for a candidate rule classifier against the backtest corpus #8085), pinned by an exact-string snapshot test.
• renderBacktestComparison renders clearly separated Regressed/Improved sections (an empty side renders (none), so an empty regression list can never read as if something regressed) and a closing verdict line; the regressed line is the pinned literal Verdict: REGRESSED — do not merge, so the follow-up CI-wiring consumer can detect the regressed case by string match without re-implementing the comparison.
• Barrel export added to packages/loopover-engine/src/index.ts on its own line immediately after the backtest-compare.js line, per the issue's deliverable.

Closes #8088

Test plan

• packages/loopover-engine/test/backtest-report.test.ts (node:test, per the issue's deliverable): exact-string snapshot of a non-null report; N/A (never 0/null/empty) for null ratios; the literal REGRESSED wording on a regressed comparison; improved comparison claims no regressed axis; unchanged comparison lists no axis on either side; byte-identical determinism for both renderers
• test/unit/backtest-report-engine.test.ts (vitest, importing the engine source): same cases — this is what exercises the .ts for codecov/patch, since @loopover/engine imports resolve to dist
• Local coverage on the new file: 6/6 lines, 8/8 branches (both null-render sides, all three verdict arms, both empty/non-empty section sides)
• npm run test --workspace @loopover/engine green; npm run typecheck green
• CI validate

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[JSONbored]

#8116 was opened first, which links to close the same issue.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.01%. Comparing base (e2d8ef8) to head (c774db9).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #8120       +/-   ##
===========================================
+ Coverage   64.35%   92.01%   +27.65%     
===========================================
  Files         759      760        +1     
  Lines       77327    77335        +8     
  Branches    23376    23380        +4     
===========================================
+ Hits        49766    71162    +21396     
+ Misses      23344     5061    -18283     
+ Partials     4217     1112     -3105     

Flag	Coverage Δ
shard-1	57.41% <0.00%> (+3.43%)	⬆️
shard-2	51.10% <0.00%> (?)
shard-3	54.45% <100.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
...loopover-engine/src/calibration/backtest-report.ts	100.00% <100.00%> (ø)

... and 438 files with indirect coverage changes