Numpy dtype support breaks between pydantic 1.7 to 1.8

[dazza-codes]

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

Numpy dtype support breaks between pydantic versions:

-version = "1.7.3"
+version = "1.8.1"

$ poetry show pydantic
name         : pydantic
version      : 1.8.1
description  : Data validation and settings management using python 3.6 type hinting

dependencies
 - python-dotenv >=0.10.4
 - typing-extensions >=3.7.4.3

$ poetry show numpy
name         : numpy
version      : 1.20.1
description  : NumPy is the fundamental package for array computing with Python.

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

$ python -c "import pydantic.utils; print(pydantic.utils.version_info())"
             pydantic version: 1.8.1
            pydantic compiled: True
                 install path: /opt/conda/envs/jupiter-gis/lib/python3.7/site-packages/pydantic
               python version: 3.7.10 | packaged by conda-forge | (default, Feb 19 2021, 16:07:37)  [GCC 9.3.0]
                     platform: Linux-4.15.0-136-generic-x86_64-with-debian-buster-sid
     optional deps. installed: ['dotenv', 'typing-extensions']

import pydantic
import numpy as np

class Bin(BaseModel):
    lower: float
    upper: float
    dtype: np.dtype = Field(default=np.dtype("uint8"))

    class Config:
        allow_mutation = False
        # this is required for numpy.dtype
        arbitrary_types_allowed = True

While loading pytest root conftest.py there is an error on import like:

pydantic/fields.py:249: in pydantic.fields.Field
    ???
pydantic/fields.py:172: in pydantic.fields.FieldInfo._validate
    ???
E   TypeError: Cannot interpret 'PydanticUndefined' as a data type