Partiality in ACSets handled in a well-defined and well-understood way

[olynch]

Currently, the story for partiality in ACSets is that we assume it doesn't exist, except for when it's convenient in building them.

There are also some subtle bugs, some of which we know exist (like int-valued attributes defaulting to zero) and some of which we are only fearful of (???) due to the janky way we currently handle partiality.

For this PR, we should have:

• A thorough control and understanding over where partiality will throw errors, where partiality will return default values, and where partiality will return "none" types.
• A coherent story for partial ACSets, where variables for attributes might be introduced
• A coherent story for ACSet transformations that could be partial.

As part of this work, acsets also have been migrated into a separate folder, and no longer depend on presentations.

[epatters]

Another clean and elegant refactor of the acsets machinery! This will open the door to all sorts of acset extensions, such as sparse or distributed acsets.