High-budget economic persona for the denial-of-wallet cohort (Launch 1, inc 4)

[MarouaBoud]

Launch 1, increment 4 (follows #25/#26/#28). Makes the embedded costbomb cohort's amplification real, not just goals assigned.

The gap this closes

Increment 2 routes the adversarial cohort to costbomb's cost-explosion goals, but stampede's heuristic brain models tokens from token_budget — so the attack didn't actually inflate cost. A dedicated high-budget persona fixes that.

What

• New economic persona in the core pack: denial_of_wallet attack, aggressive retry, patience 20, token_budget: 120000 (deliberately huge). is_adversarial → increment 2's goal-injection routes it; the high budget makes _modeled_tokens produce real amplification vs benign personas.
• Updated the two six-personas → seven-personas count tests.

Verification

New tests prove the persona is adversarial + high-budget and models more cost per turn than a frugal agent. Full suite 144 passed, 7 skipped (provider-gated), ruff clean. Built in an isolated git worktree — no effect on any in-progress branch.

Run it: population.mix: {economic: 0.1}.