I can't be bothered to look it up (sorry), but I recall reading words somewhere to the effect that if the OAT is very cold, there may be ice crystals in the air that will pass harmlessly through the engine (presumably melting in the process), UNLESS you should happen to use enough carb heat to warm the air into the icing range. Probably somewhat unusual though.
Round my parts we check downwind / base and return to cold for finals, purely 'cos under normal circumstance you get more oomph on a go-around with it cold.. UNLESS there's reason to think we are getting icing, in which case it stays hot. Carb icing can occur in relatively hot (+20C) temperatures given the right humidity etc.