Statistics of Goodness of fit measures for revealed preference tests: complexity results and algorithms