taking cosine of large integer causes panic

[jacobhenn]

replication

2<enter>1e6^c

error message

thread 'main' panicked at 'internal error: entered unreachable code: domain checks failed to detect non-finite result (NaN) in unary operation on Cos(Var("x"), Radian)', src/expr/cast.rs:81:17

[jacobhenn]

I think the culprit is that cos(inf)=NaN, and 2^10^6 is approximately inf.

This issue raises the question: why is the finiteness check on line 79 even there? Approximating 2^10^6 displays inf, and honestly what else should it display if we want to stick with floats? I think that the solution to this issue is simply to remove that panic and let NaNs be NaNs.