Move pi to trig and e to exp, fix atan2

[Izaakwltn]

There are some types that need pi or e but shouldn't need Complex or Polar in order to satisfy Elementary to get them.

This shouldn't break external code, since Elementary will be a marker class and will still contain them.

@stylewarning

I've also checked this against cl-quil/discrete and it doesn't seem to have broken anything.

[stylewarning]

Does this current rev actually type check? It seems like a type error to call BFConst on something that's not Unit -> Big-Float. See comments herein.

[stylewarning]

I'm concerned about why this program is type checking... BFConst :: (Unit -> Big-Float) -> Big-Float but you're providing it Big-Float in certain places. Are the tests passing because MPFR is being loaded (which doesn't have this same issue) and not the fallback?

[stylewarning]

Sorry for the noise. The tests are failing for Allegro. You can try testing with :coalton-portable-bigfloat in your *features* before you compile Coalton.

[Izaakwltn]

Tests are all passing now, and default Big-Float pi and ee are guaranteed to be necessary precision

[stylewarning]

a few little things but I think they ought to be addressed for standard library cleanliness