[Classification] Implement deterministic dependency, test, lint, and build rule families

[Dyu20705]

Problem statement

The first decision families should cover common Python CI failures using explicit, inspectable rules over structured evidence. Rules must prioritize precision and safe abstention over broad but unreliable coverage.

Categories

• dependency
• test
• lint_typecheck
• build_compile

Checklist

• Define stable rule ids, versions, descriptions, and priority metadata
• Implement dependency resolution/install/toolchain signatures
• Implement pytest collection/assertion/test-process signatures
• Implement lint and type-check signatures
• Implement Python build/compile/packaging signatures
• Require evidence references for every matched rule
• Produce documented rule_strength inputs rather than probability/confidence
• Add category-specific safe recommended actions
• Add positive, negative, mutation, multi-signal, and ambiguous fixtures
• Prevent repository name, branch, or incidental text from acting as hidden shortcuts

Acceptance criteria

• Each result includes category, decision_source: rule, rule id/version, score type, and evidence references
• Rule output is deterministic across repeated runs
• Positive examples match the intended family
• Negative and mutation examples protect against common false positives
• Conflicting or weak signals are passed to [Decision Policy] Implement rule priority, score semantics, conflict handling, and abstention #24 rather than forced into a category
• Rules never claim calibrated probability
• Test-process infrastructure errors are not mislabeled as test assertion failures when stronger evidence indicates another category
• Output conforms to [Core] Implement versioned input, evidence, and output schemas #34

Dependencies

• Parent epic: [Epic] Deterministic Failure Decision Engine #17
• Fixtures: [Fixtures] Build sanitized, provenance-tracked offline regression corpus #9
• Evidence parser: [Log Parsing] Extract bounded, reproducible failure windows and command context #11 and [Log Parsing] Parse annotations and filter noise without losing causal evidence #12
• Schemas: [Core] Implement versioned input, evidence, and output schemas #34
• Decision policy: [Decision Policy] Implement rule priority, score semantics, conflict handling, and abstention #24