fix(m5-scorecard): reconcile scientific-method nested replication evidence with top-level DIRECTIONAL-NOT-REPLICATED verdict

- Nested replication evidence verdict changed from ELEVATE to SIGNIFICANT-REPLICATION-SIGNAL
- replicated:true -> false (overall skill remains directional-not-replicated)
- Note rewritten: honest representation of +8.0pp p=0.001 replication signal
  that did not earn ELEVATE because primary (+5.3pp p=0.061) failed the p<0.05 gate
- MD judge evidence table: robust-ELEVATE -> directional-not-replicated
- Evidence numbers preserved; no eval re-run; no other verdict changed