It's qualitative research: as long as you take sufficient precaution picking the subjects to avoid obvious blind spots, it should be enough to get a rough idea of glaring mistakes or problems and to get a good feel of how users perceive and navigate the UI. It also allows you to individually observe effects that would often be "averaged out" by a larger sample. It's being used as a first step (before they conduct a broader quantitative research) so it sounds pretty valid.
GitLab’s UX Researcher here. TheCoreh has summed things up well but I thought I’d share my own thoughts.
During usability testing, you eventually reach a point where users start demonstrating the same behaviour. Once that happens, it’s not really worth speaking to more users, as you’re not learning anything new. You’re better to implement changes based on the user feedback you have and then re-test with different users. The navigation has been through 3 rounds of usability testing (24 users in total). Sometimes we tested with more than one prototype at once, hence us testing with a few extra users.
We conducted usability testing to get us to a point where we felt comfortable sharing the new navigation with a much wider audience. There’s still more research to come!
On the other hand, you do not need to run a scientific study to make every UX decision. There is a lot of "information" (in an abstract sense) in a person's intuition, experience and common sense. Using a small usability group to validate is quite sensible.
It's a slightly more formalised version of the 'corridor' UX test, where you grab the first person passing in the corridor and find out what isn't obvious to them about what you've just built.