Add 6-axis 0-3 scoring to meta filter with min-axis floor of 1

Extend FILTER_SYSTEM (line 19-47) prompt to require an additional `axis_scores` object: {specificity:0-3, falsifiable:0-3, solo_feasible:0-3, blast_radius:0-3, reversibility:0-3, evidence:0-3} — same six axes the proposer already self-scores (m_a_self_score). Keep the binary verdict field. In scoreProposal() after parseJsonLoose, validate axis_scores present and integer-clamped 0-3. Add rule: if min(axis_scores) < 1, force verdict='DROP' with reason `min_axis_floor: <axis>=0` even when the LLM returned KEEP. Persist axis_scores into the meta_filter_score move output JSON (already-existing output field) so downstream calibration can read them.

Of 19 historical meta proposals at filter stage, at least 1 had an axis the proposer self-scored 0 (proposer self-score data in moves table for meta_genesis). The min-axis floor would route any LLM-judge-issued 0 on the same axis to DROP, producing a measurable filter_rejected delta vs the current binary path on the same input set.

Re-run filter on the 19 historical proposed rows in dry-run with the new prompt. If the judge LLM never issues any axis_score < 1 across all rows (the floor never fires), the floor is inactive and we got an axis structure without behavior change — keep the axes for calibration, drop the floor. If >8/19 flip to DROP, the floor is too aggressive — lower threshold or kill.

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