support for built-in `dataclass` breaks pickle

[aimestereo]

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: 1.7.2
            pydantic compiled: True
                 install path: /Users/aimestereo/.pyenv/versions/3.7.6/envs/feed/lib/python3.7/site-packages/pydantic
               python version: 3.7.6 (default, Jan 30 2020, 16:07:03)  [Clang 11.0.0 (clang-1100.0.33.17)]
                     platform: Darwin-19.6.0-x86_64-i386-64bit
     optional deps. installed: ['typing-extensions']

thanks to @PrettyWood 1.7 Introduced cool support of built-in dataclass. But in our project this broke pickle

import pickle
import dataclasses

import pydantic


@dataclasses.dataclass
class BuiltInDataclass:
    value: int


class PydanticModel(pydantic.BaseModel):
    built_in_dataclass: BuiltInDataclass


print(pickle.dumps(
    PydanticModel(built_in_dataclass=BuiltInDataclass(value=0))
))

output:

---------------------------------------------------------------------------
PicklingError                             Traceback (most recent call last)
<ipython-input-14-7d67a44cf353> in <module>
     15
     16 print(pickle.dumps(
---> 17     PydanticModel(built_in_dataclass=BuiltInDataclass(value=0))
     18 ))

PicklingError: Can't pickle <class '__main__.BuiltInDataclass'>: it's not the same object as __main__.BuiltInDataclass

---

For better understanding here's an issue of pickle outside of pydantic context. This is an old pickle problem, when different classes have the same name, simple way to reproduce:

import pickle


class A: pass


# this is an analogy of how pydantic convert built-in dataclass
B = type(A.__name__, (A,), {})


print(A)
# <class '__main__.A'>

print(B)
# <class '__main__.A'>

pickle.dumps(A)
# b'\x80\x03c__main__\nA\nq\x00.'

pickle.dumps(B)
# error

output

---------------------------------------------------------------------------
PicklingError                             Traceback (most recent call last)
<ipython-input-2-b97d2c67736c> in <module>
     10
     11 print(pickle.dumps(A))
---> 12 print(pickle.dumps(B))

PicklingError: Can't pickle <class '__main__.A'>: it's not the same object as __main__.A

---

As a result we can't use this feature, we're stick to intermediate classes that converts buit-in dataclasses to pydantic:

class SerializedDataclass(pydantic.BaseModel):
	# copy/paste of all fields from built-in dataclass
    value: float