Disable promotion from Float64 to BigFloat

[araujoms]

You can just redefine the constructor:

function BigFloat(x::Float64)
    throw(error("Imprecise."))
end