feat: replace binary test_first_indicator with three-way test classif…

[kenjudy]

…ication

Resolves #25. The binary test_first_indicator collapsed three meaningfully different commit types into one metric, making large prod-only commits (the clearest AI drift signal) indistinguishable from test-only commits (a positive TDD signal).

Replaces test_first_pct with three granular summary fields:

• test_coverage_rate: commits with test AND prod files / total
• test_isolation_rate: commits with test-only files / total
• uncovered_prod_rate: large prod-only commits (no tests) / total

Per-commit fields added: test_only_commit, uncovered_prod_commit. classifyDoraArchetype updated to use uncovered_prod_rate as the foundational-challenges signal. generateInsights emits targeted warnings and positive signals for each rate. Both GitHub Actions workflows mirrored. metrics-specification.md updated throughout.

Also skips pr-metrics.yml job for dependabot[bot] — Dependabot PRs run with a read-only token; the createComment call always fails with 403 and dependency bumps have no analytical value.

133 tests passing, lint clean, typecheck clean.

[github-actions]

PR Analysis

Size: medium (based on production code)
Production Code: 135 lines (8 files)
Test Code: 136 lines (4 files)
Total: 271 lines (12 files)
Test-to-Production Ratio: 1.01:1

Concerns

• 1/1 commits exceed 100 production lines
• 1/1 commits touch more than 5 files
• Median net additions ratio 0.62 exceeds 0.50 - possible batch-acceptance pattern

Strengths

• Strong test-first discipline across commits
• Message quality 100% meets discipline threshold

Commit Analysis

Total Commits: 1
Average Commit Size: 135 production lines
Average Files per Commit: 12.0

Metric	Value
Large commits (>100 prod lines)	1/1 (100%)
Sprawling commits (>5 files)	1/1 (100%)
Test-first discipline	1/1 (100%)
Message quality	1/1 (100%)
Median net additions ratio	0.62
Test-only commits	0
Production-only commits	0

Commit Details

2454a62 Ken Judy (4/7/2026)
feat: replace binary test_first_indicator with three-way test classification
135 prod lines, 12 files [large, sprawling, test+prod]

Test Coverage

Test Adequacy: good

• Good balance of production and test code

Target ratio: 0.5-2.0 test lines per production line

DORA Capability Assessment

Archetype: legacy-bottleneck
High sprawl combined with large commits. AI may be making cross-cutting changes worse.

Capability	Metric	Value	Target
Small Batches	Large commit %	100%	<20%
Small Batches	Sprawling commit %	100%	<10%
Version Control	Test-first discipline	100%	>50%
Version Control	Message quality	100%	>60%
AI Risk Signal	Net additions ratio (median)	0.62	<0.50
Automated by Code Metrics Workflow