warn in SparseMatrixExp constructor that the type is sparse

[mforets]

the lazy matrix exponential is only defined for sparse matrices; we could have a better error message than:

julia> A = SparseMatrixExp(rand(2, 2))
ERROR: MethodError: Cannot `convert` an object of type Array{Float64,2} to an object of type LazySets.SparseMatrixExp
This may have arisen from a call to the constructor LazySets.SparseMatrixExp(...),
since type constructors fall back to convert methods.
Stacktrace:
 [1] LazySets.SparseMatrixExp(::Array{Float64,2}) at ./sysimg.jl:77

something along the lines:

SparseMatrixExp(M::Matrix{N}) where {N} =
        error("only sparse matrices can be used to create a `SparseMatrixExp`")

an example in the docstring would also help.