Add v2_a11 evidence-emergence axis to filter_score

Insert new axis v2_a11 ('evidence_emergence_shape') after v2_a10 in the axis array. Score 0-3 based on whether the hypothesis explicitly predicts a future observable evidence event with a bounded date window: 0 = no observable event named; 1 = event named but no deadline; 2 = event + deadline + ambiguous observer; 3 = event + deadline + named observer/data source. Add axis weight constant matching A1-A10 baseline. Update the composite-score sum to include A11. Add axis to the axes-rubric block in the LLM scoring prompt and to the JSON schema the prompt expects back.

Forecaster-shaped candidates (per S183 pivot) score 2-3 on A11; legacy product-shape candidates (e.g. SENTINEL-retread, 847f7e-shape) score 0-1. Composite-score spread between shapes widens by 2-3 points, making the forecaster lane separable at the filter stage rather than at council.

Rerun A11 over the 43 graduated candidates in engine.db plus the last 100 filter_rejected. If A11 score distribution shows no rank correlation (Spearman <0.3) with the S157 NBJ sweep's ROBUST vs FRAGILE classification, the axis is measuring noise, not shape — remove it.

Rejected at the council verdict. The two-judge council did not find the case strong enough to advance to Commander review.