Right now we get the tour and privacy settings modals every time we run the tests.

Our "time to first query" is also encountering the modals so the time includes that. We probably want to measure the typical case where you don't get the modals. Might as well measure with and without, though.