That is not the hallmark of literate programming, but the hallmark bad/unhelpful code comments or docs. Like, you can also have bad/unhelpful function names or unclean and tangled implementations.
In Ruby, tests are the same you'd write in any other language. You test that things do what they are supposed to do, not the types of parameters or return values.
The Atmos does not have a seconds hand. All models except two do not have minute indications.
I believe that is going to difficult measuring accuracy. You can approximate something if you compare it over a long period of time, perhaps at 12 o'clock, or as much 12 o'clock as you can tell by inspecting the hands alignment.
"TeX: The Program" is a joy to read.