Run v2 data: every trait-specific neuron

runs/v2 (2,230 neurons: v1 + 2,143 trait-specific + 250 random comparison neurons) with measured properties and automated investigations. Extract in the repo root and run: python explorer/server.py --run runs/v2 --no-model