Need to widen to <:Number to work with Dual number auto diff

[Moelf]

LorentzVectors.jl/src/LorentzVectors.jl

Line 35 in 14ebb61

struct SpatialVector{T <: AbstractFloat}

[JLTastet]

Thanks for the suggestion! That indeed sounds like a good use case.

However, changing AbstractFloat to Number could allow LorentzVectors to be constructed from integers. Do you know if there is an abstract type for real numbers?

Alternatively, we could add a special constructor for integers, that throws an error when called.
EDIT: I am already doing that with a converting constructor. Actually, maybe some of the changes from my loose-typing and complex branches could be merged. There is actually a strong use case for using non-scalar types like (Pauli) matrices as the element type of a LorentzVector, as long as this type is closed under a Lorentz transformation.

[Moelf]

I wouldn't worry too much about that aspect; programmers need to know the computer number types and their limitations when they do it anyway.