On a test of electromagnetic T-violation in charged K decays