Avoid generic fallback for Ac_mul_B

[mlubin]

Writing X'*y can lead to performance warnings. Looks like it's using the generic fallback.

julia> m = Model();

julia> @defVar(m, X[1:3,1:3]);

julia> @defVar(m, y[1:3]);

julia> @which X*y
*[...] at /home/mlubin/.julia/v0.4/JuMP/src/operators.jl:390

julia> @which X'*y
Ac_mul_B{T,S}(A::Union{DenseArray{T,2},SubArray{T,2,A<:DenseArray{T,N},I<:Tuple{Vararg{Union{Colon,Int64,Range{Int64}}}},LD}}, x::Union{DenseArray{S,1},SubArray{S,1,A<:DenseArray{T,N},I<:Tuple{Vararg{Union{Colon,Int64,Range{Int64}}}},LD}}) at linalg/matmul.jl:119

[joehuchette]

I really didn't want to get in the business of overloading A_mul_B and friends, but maybe it's inevitable.

[mlubin]

There are likely still performance issues in the overloads, but this is stale because all of the linear algebra has been rewritten.