Assessing estimated velocity-depth models: Finding error bars in tomographic inversion