Adversarial Tradeoffs in Linear Inverse Problems and Robust State Estimation