equality for character types: either let it error out, or implement it properly

[fingolfin]

Right now == for generic character (aka character types) is the default Julia one, i.e. using identity ===. This may confuse some users. I think it would be better if this threw a helpful error message.

Even better would be a "proper" equality test, but this might be non-trivial. E.g. arguably these two are equal, but deciding this requires a non-identity re-mapping of the parameters

julia> x = linear_combination([2,1], [T[4],T[4]])

julia> y = linear_combination([1,2], [T[4],T[4]])

[SoongNoonien]

Should the simple character types be treated the same here? I think all the nice things mentioned in #214 and the equality could probably be implemented very easily for the simple characters.

[fingolfin]

In which sense do you consider treating them differently? Perhaps your idea is that we could implement a functioning == for simple types rather simply -- so yeah, I'd be fine with having == do something useful for simple types even if it were mapped to error for the "full" types at the same time.