I agree with you in part - I find the biological explanations very unsatisfying given the many peculiar things we are socialized to do (proms, engagement proposals, baby showers, amongst many others, likely do not find their origins in our genes).

An analogy I thought of while reading this was that of a broken random number generator function. We're more likely to assess the software/environment in which the code is running than to blame the underlying computer.

This doesn't change my overall conclusions, but I think the role of work colleagues in personal friendships is field-dependent, and even then, workplace dependent. In my industry, some work groups spend inordinate amounts of time together, whereas I would rather go home at the end of a work day. These differences don't map cleanly to gender composition.