Junior college: I suppose a lot can depend on the specifics, but as I understand it most of these won't offer the advanced coursework that graduate programs are interested in. So doing this could in principle boost your GPA. But I think you'd be better off returning to a full university for an additional year if this is your strategy.
Certificate programs: similar to above. I think one way where this *might* be of some benefit is if, say you were to go through something like a machine learning boot camp, and then seek to get into some kind of a PhD project that would be using the machine learning to help solve some problem in physics. Sometimes project-specific skills can make a difference in admissions.
University reputation: Personally, I think this is one factor that tends to get exaggerated somewhat. One of the big reasons for the GRE is to help mitigate school-to-school differences. I think university reputation might be more of a factor in the extreme cases. MIT grads, for example, would probably be seen in a better light than graduates from a small school who hasn't had any graduates go on to graduate school. But in most cases they're not going to scale your grades. It's best not to count on your school's reputation for anything.
Who you know: This can actually be a pretty big factor if for no other reason than the fact that a good mentor with a solid knowledge of the field who knows your skills and strengths can give you advice about which programs to apply to. Sometimes as a graduating student it can feel as though you're taking shots in the dark at which programs to apply to. Insider knowledge can help you identify programs that are taking on more students, or might be looking for someone with any unique skills that you bring to the table. Another factor at play is that not all professors are equal. For one, if the person writing your reference letter is well known by the people reading the reference, it will tend to carry more weight. Also, not everyone knows how to write a decent reference letter in the first place.
It might also help to convert these factors into some kind of a standard currency. If, for the sake of argument, you define GPA as your standard currency, and your actual GPA is a 3.3. Things like having a full peer reviewed publication might bump you up to an equivalent of a 3.4, maybe a 3.5 if your the first author and your mentors have letters that state you did all the work yourself. A reference letter might be able to affect this too, but remember just about all applicants are going to have a least 'good' reference letters. So this might push you up the scale by 0.1 or so. Maybe that gives you a ballpark to play in.