I'd object
Step back for a moment and think about: What's the logic to limit the number of samples between bracketing standards? The sense is to have a specific scheme for recalibration. Think of it as the maximum time (!) which is allowed between reference injections. In this sense, of course blank injections would also count to the number of samples. Otherwise it would be perfectly allowed to have, let's say, 50 blank injections and 6 samples between your bracketing standards.
On the other hand, of course you can just take the method description literally: A sample is a sample and a blank is no sample. Just considering this method description, you would be perfectly safe to view the blanks NOT as samples.
Just another case why GMP is not the same as scientifically sound.
If I were in your situation, I'd look at the specific method: Are you confident (based on past experience) that the method (including samples and references) is stable enough to tolerate 6 samples plus blank injections between the reference injections? If yes, I'd run it like that.