This is anecdata, but with my own very limited (by choice) website analytics, I see a strong correlation between Linux users and headless browsers. So while my Linux user base seems higher than ever before so is my headless traffic. When I remove headless traffic, my Linux user base is in that 2-3.x% region.