Add support for dataclass fields init

[dmontagu]

Companion to pydantic/pydantic-core#1163 to support init=False for dataclass fields.

Still needs tests.

I checked that this PR, combined with the change to pydantic core linked above, was working by doing:

from dataclasses import field

from pydantic.dataclasses import dataclass

@dataclass(config=dict(extra='forbid'))
class A:
    x: int
    y: int = field(init=False)

    def __post_init__(self):
        self.y = self.x + 1

print(A(x=2, y=4))

The reason I think it's important to track init on the schema and not just ignore fields with init=False completely is for the sake of validate_assignment. However, I will note that I believe there are still some bugs/unimplemented aspects of validate_assignment for dataclasses, as you can see by looking for TODO comments in pydantic (e.g., here and here). So I wouldn't be surprised if validate_assignment doesn't work perfectly out of the box, but I think this is the right approach to implementation to make it feasible to get that working in the near future.

Fix #8454

Selected Reviewer: @samuelcolvin

[codspeed-hq]

CodSpeed Performance Report

Merging #8552 will not alter performance

_{Comparing support-dataclass-init (fec8cdb) with main (8e9b107)}

Summary

✅ 10 untouched benchmarks

[cloudflare-pages]

Deploying with    Cloudflare Pages

Latest commit:	fec8cdb
Status:	✅  Deploy successful!
Preview URL:	https://a6013309.pydantic-docs2.pages.dev
Branch Preview URL:	https://support-dataclass-init.pydantic-docs2.pages.dev

View logs

[sydney-runkle]

🤷‍♀️ CI expected to fail until we get the new minor release of pydantic-core out (tomorrow) and pull that in. Will leave this in draft state until then.

[sydney-runkle]

Please review

[samuelcolvin]

LGTM, what happened if you used init=False before now?

Is there any chance some existing code breaks?

[sydney-runkle]

It didn't do anything 🤷‍♀️. I think code might break if you had init=False specified before, and now it actually works, but I don't think that's a problem.

[dmontagu]

Yeah I guess it could be considered a bugfix if that helps justify the change

[samuelcolvin]

Sounds good, makes sense.