Add common stats collection objects

[will-langdale]

We need a more fine-grained method to collect DAG statistics, and a common method to log new ones. This introduces it.

Fixes #419

🛠️ Changes proposed in this pull request

Adds a maximally extensible and flexible statistics object who can provide methods that are both context managers or decorators, as needed.

The way this is achieved is:

• The DAGStats.time decorator creates a _StatLogger template. This is refreshed on each _StatLogger.__call__().
• When _StatLogger is a decorator, name and DAGStats are inferred by introspecting self, and it logs against them. If they don't exist, it'll just print logs. Because of this it can work fine on standalone functions as well as DAGs
• When _StatLogger is a context manager, there's no self to introspect, so you pass in the name and stats instance manually if you need 'em

👀 Guidance to review

In tests, the patching pattern is non-standard for this repo, because I think it creates a significantly more concise and readable set of tests in this instance.

update.py doesn't need the full logger object, so instead we use the adapter existing memory logging function. This function would also, one day, be perfect for _StatLogger if we wish.

🤖 AI declaration

AI written, human reviewed.

✅ Checklist:

• This is the smallest, simplest solution to the problem
• I've read our code standards and this code follows them
• All new code is tested
• I've updated all relevant documentation (select all that apply)
  • API documentation (docstrings and indexes)
  • Tutorials
  • Developer docs