`SquareEye` multiplication not specialized

Similar issue: #107

julia> A = rand(1000,1000);

julia> B=Eye(1000);

julia> @btime A*B;
  1.414 ms (3 allocations: 7.63 MiB)

julia> @btime copy(A)
  1.097 ms (3 allocations: 7.63 MiB)

Downstream issue:

julia> using ClassicalOrthogonalPolynomials, ToeplitzMatrices

julia> B=Toeplitz(1:∞,1:∞)
ℵ₀×ℵ₀ Toeplitz{Int64, InfiniteArrays.InfUnitRange{Int64}, InfiniteArrays.InfUnitRange{Int64}} with indices OneToInf()×OneToInf():
  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  …  
  2  1  2  3  4  5  6  7  8   9  10  11  12  13  14  15  16  17  18  19  20  21  22
  3  2  1  2  3  4  5  6  7   8   9  10  11  12  13  14  15  16  17  18  19  20  21
  4  3  2  1  2  3  4  5  6   7   8   9  10  11  12  13  14  15  16  17  18  19  20
  5  4  3  2  1  2  3  4  5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
  6  5  4  3  2  1  2  3  4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  …
  7  6  5  4  3  2  1  2  3   4   5   6   7   8   9  10  11  12  13  14  15  16  17
  8  7  6  5  4  3  2  1  2   3   4   5   6   7   8   9  10  11  12  13  14  15  16
  9  8  7  6  5  4  3  2  1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
 10  9  8  7  6  5  4  3  2   1   2   3   4   5   6   7   8   9  10  11  12  13  14     
  ⋮              ⋮                ⋮                   ⋮                   ⋮          ⋱

julia> S=Legendre()
Legendre()

julia> S\(S*B) # freeze

Start by reproducing the Julia benchmarks for AB versus copy(A) with SquareEye, then inspect the existing multiplication path for SquareEye. Also test the downstream S(SB) example with ClassicalOrthogonalPolynomials and ToeplitzMatrices; done means the specialized multiplication removes the reported overhead and supports that use case.