effective_disapproval_integrity and effective_endorser_min_integrity duplicated an identical 14-line parse-and-warn body. The promotion/demotion label pipeline (apply_post_integrity_adjustments) also lacked end-to-end test coverage, meaning accidental removal of either adjustment step would go undetected.

Changes

• Extract resolve_integrity_level helper — consolidates the shared "empty → default, unrecognised → warn + default" pattern into a single function; both callers now delegate in one line each:

fn resolve_integrity_level(value: &str, default: MinIntegrity, caller: &str) -> &'static str {
    let v = value.trim();
    if v.is_empty() {
        default.as_str()
    } else {
        MinIntegrity::from_policy_str(v)
            .unwrap_or_else(|| {
                crate::log_warn(&format!("{}: unrecognised level {:?}, defaulting to '{}'",
                    caller, v, default.as_str()));
                default
            })
            .as_str()
    }
}

fn effective_disapproval_integrity(ctx: &PolicyContext) -> &'static str {
    resolve_integrity_level(&ctx.disapproval_integrity, MinIntegrity::None, "effective_disapproval_integrity")
}

• Add 4 promotion/demotion label integration tests — exercise the full issue_integrity / pr_integrity pipeline to verify:
  • promotion_label raises a NONE-association issue to writer (rank 3)
  • demotion_label caps an OWNER-association issue to none (rank 1)
  • promotion_label raises a forked PR from reader (rank 2) to writer (rank 3)
  • demotion_label caps a merged PR from merged (rank 4) to none (rank 1)