Add check for dataclass field default ordering

[darricklaidin]

This PR introduces a validation step to raise an error when a dataclass field without a default follows a field with a default, addressing issue #492.

This ensures dataclasses are declared in the correct field order, aligning with Python's rules for @dataclass.

The core change involves modifying the get_dataclass_init function when getting the dataclass synthesized fields to perform an ordering validation check (similar to #410). Due to the additional properties of a dataclass, additional checks for init, kw_only, and default have to be made.

Separately, certain test cases that I've added rely on issue #493 being fixed. For now, I have commented those out - and will uncomment and verify them once the aforementioned issue has been resolved.

[darricklaidin]

Let me know if there's a better place to put the validation logic, e.g., between lines 66-67.

[rchen152]

Thanks, this looks great overall! Just a couple minor comments.

[rchen152]

Why are these tests commented out? If they're broken, our convention is to add # E: comments matching the current behavior, plus a bug = ... line describing why the current behavior is buggy. Here's an example:

pyrefly/pyrefly/lib/test/simple.rs

Lines 1494 to 1503 in 747a4bc

	testcase!(
	bug = "Mapping.get with a default gives the wrong answer",
	test_mapping_get,
	r#"
	from typing import assert_type
	import os
	compiler = os.environ.get("CXX", "cl")
	assert_type(compiler, str) # E: assert_type(str | Any, str) failed
	"#,
	);

[rchen152]

This is a really thorough set of test cases, thank you!

[rchen152]

nit: dataclass is usually written all lower-case.

Suggested change

	"DataClass field '{}' without a default may not follow DataClass field with a default", name
	"Dataclass field '{}' without a default may not follow dataclass field with a default", name