Standardise naming of base classes

[mscroggs]

Currently, we use a mixture of naming conventions for base classes, eg:

• The base node is Node
• The base component (added in Lagrangian respects effect handlers #94) is _CPComponent

We should standardise these

[mscroggs]

I like not having an _ in node (to suggest it's not private) as I can imagine an (advanced) user wanting to define their own subclass of node. But perhaps having _ before some base classes does make sense

[mscroggs]

We could aim to all base classes BaseSomething if we think this is clearer? Although it being just Node makes for neater typing of functions that take nodes as input

[willGraham01]

I like not having an _ in node (to suggest it's not private) as I can imagine an (advanced) user wanting to define their own subclass of node. But perhaps having _ before some base classes does make sense

Yeah I used an underscore in _CPComponent because I don't envision it being something a user ever wants to use / derive something custom from directly, so it's just a useful refactoring of common functionality. The only "components" of the problem are our constraints and causal estimand, both of which are actual classes we expect the user to interact with. Likewise I agree with your Node take - it doesn't need the _ since someone may have a reason for creating their own subclass, and it's not something we want to "hide".

That being said, _CPComponent -> _BaseComponent doesn't sound too nonsensical and it probably the path of least resistance.