As a practical matter, test mating can make someone pretty sure. Not absolutely sure, but pretty close.
And test mating can certainly identify a lot of heterozygotes, so they can be removed from the breeding pool. It just cannot be certain of identifying every heterozygote.
You're right that test mating takes a lot of time and space, to raise enough pullets to laying age, so it's often not practical.
When making Easter Eggers, people often take a shortcut by using the pea comb gene: it's linked closely enough to the blue egg gene to be a useful marker. Of course the linkage can be any direction: pea/blue like Ameraucana, not-pea/blue like Cream Legbar, pea/not-blue like Brahma, not-pea/not-blue like Orpington. And crossovers happen, so you get a few right-comb/wrong-egg birds in any flock that is being selected that way.