The time/investment they spent on their two brand identity typefaces is one side of the equation, the other side is the many, many minutes spent on font decisions in each and every little project in their company if employees are allowed (or maybe even encouraged, it can be quite motivating) to dabble a bit in "visual project identity". And on font licencing awareness training, because those engineer doodles will likely be public. And a Microsoft subsidiary will be an extremely juicy target for licence vultures, github simply can't afford any mistakes in that field.
This is where the wide parameterization of the fonts comes into play: "use one of those two fonts, feel free to go wild with the parameters" is much more likely to actually be followed than "use one of those n fonts, no exceptions", for almost any value of n. And better for morale as well.
You're saying it's better for morale, but do you have any proof? My personal experience says the contrary, the company I work at did the same (albeit it's much smaller than GitHub) and I asked some coworkers what they thought about it, about three of them considered it was a waste of resources, I only remember one saying it was OK, so overall morale went down.
This is where the wide parameterization of the fonts comes into play: "use one of those two fonts, feel free to go wild with the parameters" is much more likely to actually be followed than "use one of those n fonts, no exceptions", for almost any value of n. And better for morale as well.