Consider `import-linter` or another way to document the overall architecture

For new contributors, having some documentation of the overall pins-python package structure could be helpful.

In particular, it is helpful to get an idea of the dependency relationships between modules. import-linter allows you to declare and enforce a layered architecture; for example for pins-python, the configuration could be declared as:

[tool.importlinter]
root_packages = ["pins"]

[[tool.importlinter.contracts]]
name = "pins Modular Design"
type = "layers"
layers = [
    "constructors",
    "boards",
    "drivers",
    "meta | cache | rsconnect",
    "versions | utils",
    "config | errors | _types | data",
]
containers = "pins"
exhaustive = true

The idea is that that higher layers in the list can import from lower layers in the list, but not vice versa; and modules in the same level separated by | cannot import each other either.

Alternatively, I think having a diagram or other form of documentation would be quite helpful. However, import-linter has the advantage of enforcing the structure (potentially via pre-commit); whereas documentation can become out of date.

Review the pins package structure and the proposed import-linter layers first. Done means choosing and documenting an architecture approach, with dependency relationships either enforced by import-linter and pre-commit or captured in an alternative diagram or documentation.