Fix small bug related to floating point conversion

[sglyon]

@davidanthoff, thanks for sending this.

I forgot that writing it this way would break on 0.3 (the current code is the correct syntax on 0.4).

I actually think the easiest thing to do would be just to drop the manual conversion altogether and let Julia take care of it for us.

This example works with julia 3 and 4 (on my machine):

type Foo
    x::Float64
end

x = 3
println(typeof(x))  # Int64
Foo(x)  # Foo(3.0) -- automatic conversion to Float64

Would you mind just deleting that line?

[davidanthoff]

Done!

[sglyon]

Excellent thanks!