Allow flow parameter schema generation when dependencies are missing

[desertaxle]

This PR adds a new method of generating a parameter schema for a function so users and use prefect deploy when not all dependencies are available on the current machine.

Our current method of generating a parameter schema requires us to load the flow function itself and get the signature using the built-in inspect module. This causes issues when not all dependencies are present because loading the flow causes the entire module the flow is located in to be run and can result in Python raising an ImportError.

This PR updates this approach to allow a parameter schema given an entrypoint that points to a function with the new parameter_schema_from_entrypoint function. This new function will generate a inspect.Signature object and docstring dictionary from the source code directly and use it with the existing machinery for generating a parameter schema.

Under the hood, the new functionality will load the source code at the provided entrypoint and parse it using the ast module. A new utility called safe_load_namespace will load each module-scoped import one by one and create a namespace dictionary for downstream use. Custom class definitions will also be included in the namespace. Any import errors will be caught and logged, but will not stop execution. The created namespace is used as context when walking the AST for the given function so that custom typing can be correctly resolved. In the case that typing cannot be resolved, the generated Signature will use no annotation to prevent interrupting execution. This can result in overly broad typing if the annotations rely on external types that aren't installed at deployment time.

The diff is big, but it's mostly tests.

Closes #9512

Example

Given code stored at flow.py:

from prefect import flow
from pandas import DataFrame
from pydantic import BaseModel

class MyModel(BaseModel):
    x: int

@flow
def my_flow(input: MyModel) -> pandas.DataFrame:
    # creates the dataframe in a clever way

The following call will generate the expected parameter schema with or without panadas installed:

parameter_schema_from_entrypoint("flow.py:my_flow")

The dictionary format looks like this:

{
    "title": "Parameters",
    "type": "object",
    "properties": {
        "input": {
            "allOf": [{"$ref": "#/definitions/MyModel"}],
            "position": 0,
            "title": "input",
        }
    },
    "required": ["input"],
    "definitions": {
        "MyModel": {
            "properties": {"x": {"title": "X", "type": "integer"}},
            "required": ["x"],
            "title": "MyModel",
            "type": "object",
        }
    },
}

Checklist

• This pull request references any related issue by including "closes <link to issue>"
  • If no issue exists and your change is not a small fix, please create an issue first.
• If this pull request adds new functionality, it includes unit tests that cover the changes
• This pull request includes a label categorizing the change e.g. maintenance, fix, feature, enhancement, docs.

For documentation changes:

• This pull request includes redirect settings in netlify.toml for files that are removed or renamed.

For new functions or classes in the Python SDK:

• This pull request includes helpful docstrings.
• If a new Python file was added, this pull request contains a stub page in the Python SDK docs and an entry in mkdocs.yml navigation.

[netlify]

✅ Deploy Preview for prefect-docs-preview ready!

Name	Link
🔨 Latest commit	4482c51
🔍 Latest deploy log	https://app.netlify.com/sites/prefect-docs-preview/deploys/664276a230c16c000827893c
😎 Deploy Preview	https://deploy-preview-13315--prefect-docs-preview.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site configuration.

[desertaxle]

@serinamarie Added the docstrings that we identified as missing during our sync review.

[serinamarie]

🔥