Function compute_confidence_from_components 

pub fn compute_confidence_from_components(
    evidence: &Evidence,
    conditions: &Conditions,
    contested: bool,
    n_replicated: u32,
    n_failed: u32,
    n_partial: u32,
    causal_claim: Option<CausalClaim>,
    causal_evidence_grade: Option<CausalEvidenceGrade>,
) -> Confidence

Pure-math kernel for the frontier-epistemic confidence formula. Takes replication counts and (v0.38.1) the optional causal typing as inputs so the same math drives both the legacy scalar path (compute_confidence) and the v0.32+ Project-aware path (Project::compute_confidence_for).

Replication strength schedule: clamp(0.7 + 0.1 * n_replicated + 0.05 * n_partial - 0.10 * n_failed, 0.4, 1.0)

Floor at 0.4 keeps a single failed replication from zeroing out the computation; ceiling at 1.0 caps the bonus from accumulated successes. inconclusive outcomes do not move the score (deliberate — they represent methodological ambiguity, not evidence).

v0.38.1: a causal_consistency factor multiplies the support product. None for either field is neutral (pre-v0.38 frontiers behave identically). See causal_consistency_multiplier.