An alternative methodology for imputing missing data in trials with genotype-by-environment interaction: some new aspects