The rest, IMO, are arguments that get repeated every time this issue comes up that have barely any value. For temperature, the main argument is that people won’t understand what ½ means. Counter-argument (also weak, but not weaker than the one about ½) is that digital body-for measuring body temperatures that use Fahrenheit all seem to use a digit beyond the decimal, too (https://www.google.nl/search?q=Digital+fahrenheit+thermomete... shows 98.6 as the winning value). The argument that negative temperature are hard to understand, in turn, is offset by the fact that Fahrenheit also way to easily requires the use of 3 digit numbers for normal temperatures (both also very weak arguments)
The ‘exact division is easier’ argument holds some water, but how often do you need to do exact divisions (no, you can’t use a saw to cut a two by four into two two by twos; the saw takes away material). One could also argue we should have chosen an odd prime to give almost all divisors even footing, and to prevent requiring the invention of ‘round to odd/even’ conventions.
We're talking about customary human usage. Various fields use their own units from nm to parsecs and whether they should be divisible by 10 is not part of my discussion (sure, for both of these cases it's as good as any other divisor). In fact for daily use, many metric countries retain traditional units not just for daily purposes but even for things like land. The UK is the most familiar case for Americans, but many countries are similar.
Honestly how often do you really divide any measurement by 10? Perhaps km.