feat(conformance): publish the report-only corpus in the summary

[hhimanshu]

Closes #835.

conformance-summary.json published only the enforced corpus. The report-only corpus went to the test log and nowhere else — so anything downstream that wanted the whole picture had to hard-code figures read out of local test output. That is already happening on the marketing site, and it goes stale the moment someone fixes a gap, silently, with nothing to catch it.

It is now published as a sibling of overall:

"overall":  { "passed": 10646, "total": 10649 },
"tracked":  { "passed": 2025,  "total": 2236  }

Why not fold it into overall

Because overall is what the README badge asserts — the enforced suite passes — and folding these rows in would take that badge from 100.0% to 98.3% as a side effect of a refactor. That may well be a change worth making, but it should be a decision, not something that falls out of a plumbing PR. A consumer that wants the full picture adds the two numbers; the point here is only that both come from one file.

Say the word and I'll do the badge change as a one-line follow-up.

Why the collector, not the runner

The two paths count differently. The report-only runner prints 2055/2235; the collector gives 2025/2236, because the runner evaluates volatile formulas against a pinned clock where the collector skips them.

overall is collector-derived — its 10646/10649 matches the published badge exactly — so tracked has to be too. Publishing a runner-derived figure next to a collector-derived one would be mixing methods in a single object, and any consumer adding them would get a number that means nothing.

Verification

• Full suite green: 21 passed, 0 failed.
• overall byte-identical to before, so the badge and the published report page are untouched.
• Simulated the CI summary step against the regenerated report: overall 10646/10649, tracked 2025/2236, combined 12671/12885 = 98.34%.
• tracked is Option, and the CI step is guarded, so a report without it still produces a valid summary.

Follow-up on my side once this lands: the marketing site drops its hard-coded constant and reads this file.

🤖 Generated with Claude Code

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Test Coverage by Category

Category	Unit Tests	Google Sheets Conformance	Property Cases	Total
Array	42	552/552 ✓	1,000 (2×500)	1,594
Database	35	182/182 ✓	3,500 (7×500)	3,717
Date	369	418/418 ✓	2,500 (5×500)	3,287
Engineering	245	886/888 ⚠	5,500 (11×500)	6,633
Filter	11	81/81 ✓	4,500 (9×500)	4,592
Financial	149	1,208/1,208 ✓	2,000 (4×500)	3,357
Info	0	256/256 ✓	4,500 (9×500)	4,756
Logical	121	267/267 ✓	3,500 (7×500)	3,888
Lookup	69	393/393 ✓	1,000 (2×500)	1,462
Math	536	2,006/2,006 ✓	8,000 (16×500)	10,542
Operator	87	251/251 ✓	7,500 (15×500)	7,838
Parser	83	93/93 ✓	4,000 (8×500)	4,176
Query	37	—	—	37
Statistical	516	3,191/3,191 ✓	5,000 (10×500)	8,707
Text	309	803/804 ⚠	4,000 (8×500)	5,113
Timezone	47	—	—	47
Volatile	0	—	3,500 (7×500)	3,500
Web	29	59/59 ✓	6,000 (12×500)	6,088
Total	2,941	10,646/10,649	66,000 (132×500)	~79,590

_{✓ = 100% passing · ⚠ = known deviation · The ~79,590 total counts formula evaluations (each conformance row and each property case = 1). GitHub Checks reports 3,797 Rust test functions: 2,941 unit + 159 property functions (shown as cases above) + 697 conformance/integration.}