Add v2_a11 distribution_specificity axis to filter_score.js

Append v2_a11 after v2_a10. Scored 0-3: 0 = no channel named, 1 = generic channel ('LinkedIn outreach'), 2 = named channel + audience descriptor ('UK recruitment ops Slack groups'), 3 = named channel + audience + first-contact concrete step ('DM 20 named UK recruitment ops managers w/ Late-Payer Diagnostic link'). Pull score from the same proposal text the existing v2_a7 distribution_reachability reads; do NOT re-LLM. Use a regex+token classifier: count named entities (channels, audience qualifiers, action verbs in imperative mood). Sum into composite with weight matching v2_a7 (do not double-count: clamp v2_a7 + v2_a11 contribution to <= 2x single-axis weight).

Proposals currently scoring high on v2_a7 via vague 'reach via social' phrasing will drop 1-2 composite points; proposals with named channel + audience will be unchanged. Top-10 ranking should reshuffle by ~30%, with B_PILOT-shaped (UK recruitment, named channel) proposals climbing.

Rerun on the last 50 graduated candidates. If v2_a11 score distribution has Pearson r > 0.7 with v2_a7, the axis is redundant — kill it. If <30% of graduated candidates score >=2, the rubric is mis-calibrated against current corpus and needs threshold relaxation before merge.

Rejected by filter_score. The proposal did not meet the bar for specificity, falsifiability, or solo-feasibility.