Query counts already happen at the model and Python API test layers.
Having query counts in the HTTP API tests makes things more brittle
because that includes middleware that this code should not be aware
of. We don't want to have to revisit these tests every time
another piece of middleware is added or removed.