Issue with BaseModel if used with importlib

[vfdev-5]

Checks

• I added a descriptive title to this issue
• I have searched (google, github) for similar issues and couldn't find anything
• I have read and followed the docs and still think this is a bug

Bug

Output of python -c "import pydantic.utils; print(pydantic.utils.version_info())":

             pydantic version: 1.7.3
            pydantic compiled: True
                 install path: /opt/conda/lib/python3.8/site-packages/pydantic
               python version: 3.8.2 (default, Mar 25 2020, 17:03:02)  [GCC 7.3.0]
                     platform: Linux-4.15.0-130-generic-x86_64-with-glibc2.10
     optional deps. installed: ['typing-extensions']

If I'm trying to load a module using importlib where a BaseModel is used then execution is crashed with the following traceback:

Traceback (most recent call last):
  File "loader.py", line 13, in <module>
    m = load_module("module_to_load.py")
  File "loader.py", line 7, in load_module
    spec.loader.exec_module(module)  # type: ignore[union-attr]
  File "<frozen importlib._bootstrap_external>", line 783, in exec_module
  File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
  File "module_to_load.py", line 4, in <module>
    class User(BaseModel):
  File "pydantic/main.py", line 274, in pydantic.main.ModelMetaclass.__new__
  File "pydantic/typing.py", line 293, in pydantic.typing.resolve_annotations
    def __copy__(self):

How to repro:

1. create loader.py file with the following content

import importlib.util

def load_module(fpath):
    spec = importlib.util.spec_from_file_location("mymodule", fpath)
    module = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(module)  # type: ignore[union-attr]
    return module

if __name__ == "__main__":    
    m = load_module("module_to_load.py")
    print(m)
    m.foo()

2. create module_to_load.py :

from pydantic import BaseModel

class User(BaseModel):
    id: int
    name = 'Jane Doe'

def foo():
    print(User(id=12, name="test"))

3. Run python loader.py

As use-case may seem strange, I'm a bit hestiating between a bug and wrong usage, but in the latter case, error message is unclear at all. Important part is that I have to import the module in this way and not in any other more standard python way in which pydantic indeed work without any issue.
Thanks

EDIT:

Indeed, if we add sys.modules["mymodule"] = module to load_module as suggested here, there is no issue, anymore.
I have to figure out what would be the impact of doing that for my goal.

[samuelcolvin]

I've created a PR to make resolve_annotations more lenient, see #2366.

But actually the fix for this is to use a simplified approach to loading the module, just

from importlib.machinery import SourceFileLoader

def load_module(fpath):
    return SourceFileLoader('mymodule', fpath).load_module()

...

Works fine.

[vfdev-5]

Thanks for the suggestion and for the code update !