Throw a `DimensionMismatch` error instead of `MethodError` when using wrong differentiation method for function

I find I frequently mix up derivative(), gradient() and jacobian(), especially when I'm debugging and testing several different functions and I forget to switch to the correct one. If I actually use gradient on a vector-valued function, for example, I get this:

MethodError: no method matching extract_gradient!(::Type{ForwardDiff.Tag{getfield(Main, Symbol("##58#59")),Float64}}, ::Array{Array{Dual{ForwardDiff.Tag{getfield(Main, Symbol("##58#59")),Float64},Float64,3},1},1}, ::Array{Dual{ForwardDiff.Tag{getfield(Main, Symbol("##58#59")),Float64},Float64,3},1})
Closest candidates are:
  extract_gradient!(::Type{T}, ::AbstractArray, !Matched::Dual) where T at /home/jlumpe/.julia/packages/ForwardDiff/okZnq/src/gradient.jl:76
  extract_gradient!(::Type{T}, ::AbstractArray, !Matched::Real) where T at /home/jlumpe/.julia/packages/ForwardDiff/okZnq/src/gradient.jl:75
  extract_gradient!(::Type{T}, !Matched::DiffResults.DiffResult, !Matched::Dual) where T at /home/jlumpe/.julia/packages/ForwardDiff/okZnq/src/gradient.jl:70
  ...

Stacktrace:
 [1] vector_mode_gradient(::getfield(Main, Symbol("##58#59")), ::Array{Float64,1}, ::ForwardDiff.GradientConfig{ForwardDiff.Tag{getfield(Main, Symbol("##58#59")),Float64},Float64,3,Array{Dual{ForwardDiff.Tag{getfield(Main, Symbol("##58#59")),Float64},Float64,3},1}}) at /home/jlumpe/.julia/packages/ForwardDiff/okZnq/src/gradient.jl:98
 [2] gradient(::Function, ::Array{Float64,1}, ::ForwardDiff.GradientConfig{ForwardDiff.Tag{getfield(Main, Symbol("##58#59")),Float64},Float64,3,Array{Dual{ForwardDiff.Tag{getfield(Main, Symbol("##58#59")),Float64},Float64,3},1}}, ::Val{true}) at /home/jlumpe/.julia/packages/ForwardDiff/okZnq/src/gradient.jl:17
 [3] gradient(::Function, ::Array{Float64,1}, ::ForwardDiff.GradientConfig{ForwardDiff.Tag{getfield(Main, Symbol("##58#59")),Float64},Float64,3,Array{Dual{ForwardDiff.Tag{getfield(Main, Symbol("##58#59")),Float64},Float64,3},1}}) at /home/jlumpe/.julia/packages/ForwardDiff/okZnq/src/gradient.jl:15 (repeats 2 times)
 [4] top-level scope at In[60]:1

This error is rather cryptic and it's not immediately obvious what your mistake was when you run into it for the first time. It's not a big deal, but it might be nice to overload extract_gradient!() and others to throw a DimensionMismatch or something with an informative message instead.

Start in gradient.jl at the extract_gradient! overloads and vector_mode_gradient shown in the stack trace, then reproduce the example using gradient on a vector-valued function. Done means the misuse raises an informative DimensionMismatch instead of a cryptic MethodError.