By "pure" do you mean homozygous? In which case, per your later posts homozygous X normal does produce 100% hets just looking at the genotypes for the breeding results. You can then come along afterward and apply the mutation type and figure out that for a dominant mutation the hets look like their homozygous parent but have a different genotype so will have different breeding results when they are grown up and bred.
Het X normal would however produce eggs with a 50% chance of being hets and a 50% chance of being normal regardless of the mutation type but being a dominant type the hets would show.
The belief that "het" only applies to normal looking recessive hets I think goes back to the recessive snake morphs being where most of us first learned the term het. Outside of snakes, het is understood to mean having an unmatched pair of genes and doesn't necessarily tell you what the animal looks like (without knowing the mutation type) but is very useful in understanding its breeding results. By knowing that it has two different types to choose from you can see that each offspring has a 50/50 chance as to which version of the gene (in our case mutant or normal) it gets.