Transformations between acsets involving only combinatorial data

[kris-brown]

By considering CSetTransformations between ACSets (where naturality is not demanded on attributes) we can obtain two sorts of practical constructions:

1. A kind of "labeled CSet" where we can think of the attributes as functioning as labels
2. A kind of limit construction for ACSets which acts purely on the combinatorial data. The apex can put whatever it wants for the attributes, but we can deliberately pick a representative of the isomorphism class to make it correspond to something meaningful (with valid ACSetTransformations), which Kevin has a nice categorical characterization of in an AlgebraicJulia blog post.

Implementation-wise, this means we want an abstract type of ACSetMorphism which is implemented by ACSetTransformation and CSetTransformation (therefore, this is changing the meaning of CSetTransformation, which before was an ACSetTransformation that happened to be on a schema with no attributes). There is probably a lot of functionality which is written for ACSetTransformation which would straightforwardly work for ACSetMorphism, but the minimal required functionality (e.g. dom, compose) for getting limits to work is all that is generalized for now.

It may be the case that we want something in between, where a subset of the attributes are marked to be ignored. This would give us a kind of "LACSets" - but we would then lose the ability to do #2 above, which I see as the real value of this.

Edit: I also addressed a tiny bug in the HomSearch code where monic and iso fast failure constraints were not being checked for the combinatorial Attr data.

[epatters]

Very nice! I just left a few small comments.

[epatters]

Thanks Kris!