Some * fails because of eltype

[hyrodium]

Sample code to produce the issue

julia> r = AngleAxis(1,1,0,0)
3×3 AngleAxis{Int64} with indices SOneTo(3)×SOneTo(3)(1, 1, 0, 0):
 1.0  0.0        0.0
 0.0  0.540302  -0.841471
 0.0  0.841471   0.540302

julia> eltype(r)
Int64

julia> r*[1,2,3]
ERROR: InexactError: Int64(-1.44380834268741)
Stacktrace:
 [1] Int64
   @ ./float.jl:723 [inlined]
 [2] convert
   @ ./number.jl:7 [inlined]
 [3] macro expansion
   @ ~/.julia/packages/StaticArrays/vxjOO/src/util.jl:11 [inlined]
 [4] convert_ntuple
   @ ~/.julia/packages/StaticArrays/vxjOO/src/util.jl:7 [inlined]
 [5] StaticArrays.SVector{3, Int64}(x::Tuple{Float64, Float64, Float64})
   @ StaticArrays ~/.julia/packages/StaticArrays/vxjOO/src/SArray.jl:28
 [6] macro expansion
   @ ~/.julia/packages/StaticArrays/vxjOO/src/matrix_multiply.jl:85 [inlined]
 [7] _mul
   @ ~/.julia/packages/StaticArrays/vxjOO/src/matrix_multiply.jl:67 [inlined]
 [8] *(A::AngleAxis{Int64}, B::Vector{Int64})
   @ StaticArrays ~/.julia/packages/StaticArrays/vxjOO/src/matrix_multiply.jl:8
 [9] top-level scope
   @ REPL[12]:1

How to fix this

#132 is related, but I think it's better to distinguish the type of parameters of rotation and the type of matrix.

[andyferris]

Hmm, I kinda feel if the input type is Integer we should just call float on it?

[andyferris]

If someone creates AngleAxis{Int} explicitly and it fails, it's not a big problem, it's just not supported. But if someone goes AngleAxis(...) we should give them a compatible type by default. (We need to maintain compatibility with dual numbers, etc).

[hyrodium]

Hmm, I kinda feel if the input type is Integer we should just call float on it?

Okay, I'll open a PR to do that.