Fix index dimension mismatch in TensorNetwork construction

[Todorbsc]

Closes #155

[Todorbsc]

@jofrevalles @mofeing the dimension mismatch seems to be fixed with the verification I implemented in TensorNetwork construction. I "uncommented" the TN dimension mismatch construction test and now it raises the proper exception as well as I tested the Chain example that Jofre commented in the issue, what raises an error now:

julia> mps = Chain(State(), Open(), [rand(2, 3), rand(2, 6, 4), rand(2, 5)])
ERROR: DimensionMismatch: Index D has inconsistent dimension: [3, 6]
Stacktrace:
 [1] TensorNetwork(tensors::Vector{Tensor{Float64, N, A} where {N, A<:AbstractArray{Float64, N}}})
   @ Tenet ~/QuanticGit/Tenet.jl/src/TensorNetwork.jl:58
 [2] Chain(::State, boundary::Open, arrays::Vector{Array{Float64}}; order::Tuple{Symbol, Symbol, Symbol})
   @ Tenet ~/QuanticGit/Tenet.jl/src/Ansatz/Chain.jl:99
 [3] Chain(::State, boundary::Open, arrays::Vector{Array{Float64}})
   @ Tenet ~/QuanticGit/Tenet.jl/src/Ansatz/Chain.jl:63
 [4] top-level scope
   @ REPL[3]:1

Also, let me know if there's something else needed, specially with what you commented in the issue about the unsafe_region.

[jofrevalles]

Okay, nice! Since we already have tests for @unsafe_region, and these are passing, we can merge this!

[mofeing]

Do you mind benchmarking with and without the safe guards / checks for some large TNs? Like 100-1000 tensors