Add Form-specific constructors for MPS

[jofrevalles]

This PR adds Form-specific constructors for the MPS struct, which allows us to create Matrix Product States with an specific canonical Form directly on the constructor. I believe this is the proper way to go, and we added a check_canonical_form kwarg that, if true (default), it will check if the MPS fulfills the canonical form.

TO DO:

• Regroup repeated code into an internal function. Maybe it can be named __tn_mps_from_arrays? @mofeing I can't find a good name. Group code by using NonCanonical constructor.

[mofeing]

Great! Thanks! In general everything is great. Just some suggestions for the generic MPS constructors.

Regroup repeated code into an internal function. Maybe it can be named __tn_mps_from_arrays?

You no longer need to prefix internal functions with __ since we'll soon start annotating public API with the new public keyword (well, with the @public macro from Compat.jl).

Also, what I would recommend you is not to refactor it to an internal function, but to leave that code in the NonCanonical constructor. Then Canonical and MixedCanonical would call the NonCanonical constructor and modify it.

• MixedCanonical calls the NonCanonical constructor and then modifies the form field.
• Canonical calls the NonCanonical constructor, tensorizes the alphas by requesting bond index name with inds(tn; bond=...), push them to the TN and finally modify the form field.

An example of this pattern can be found in simple_update! for MixedCanonical: it first calls mixed_canonize! to move the orthogonality center to the target sites and then calls NonCanonical version of simple_update!.

[jofrevalles]

Okay, everything should be ready now @mofeing. I separated the checks of the form in two check_form functions that are dispatched over the Form. I think this is a good idea since we would want to use these functions somewhere outside the constructor. In this case, maybe it's even better that they do not throw an error but just return false? I don't know, for me this is still fine.

[mofeing]

Great! Just some suggestions and we're done.

[mofeing]

do you need to contract to check? that can be a lil bit costly

@starsfordummies what do you think? is there some other way to check that the gammas are correct?

[mofeing]

I'm okay with merging it this like for now. But if there's a way to check it without contracting, then we should open an issue to implement it in the future.

[mofeing]

Thanks!