fix: Default to non nullable column for array and object

[gab23r]

I think is as been missed when merging the PR.
Or maybe it wasn't possible to implement at the time. In any case I don't see why object and array are nullable by default.

This is breaking, how do you want to proceed ?

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

This PR changes the default nullability for Object and Array columns to be non-nullable by default, addressing the concern that these types currently default to nullable.

Changes:

• Change Object(..., nullable=...) default from True to False
• Change Array(..., nullable=...) default from True to False

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
dataframely/columns/object.py	Makes Object columns non-nullable by default
dataframely/columns/array.py	Makes Array columns non-nullable by default

[Copilot]

Changing the default value of a public constructor parameter is a breaking API change that can silently alter existing schemas (call sites that omit nullable will now get NOT NULL). To make this safe and explicit, consider either (1) doing a deprecation cycle (keep default True for now and warn when nullable is omitted for Object/Array, then flip in the next major release), or (2) keep this flip but ensure the release process captures it clearly (changelog entry + major version bump per SemVer) and add migration guidance: 'set nullable=True to keep previous behavior'.

[Copilot]

This introduces the same silent-breaking change for Array. Additionally, unless the rest of the column types also default to nullable=False, this will make nullability defaults inconsistent across column classes, which is confusing for API consumers. Consider aligning defaults across column types (preferred) or documenting this special-case rationale prominently in the API docs/migration notes.

Suggested change

	nullable: bool = False,
	nullable: bool = True,

[Oliver Borchert (borchero)]

Thanks for the fix, this was surely an oversight! Can you still fix the tests?

I think we'll have to mark that as non-breaking as I don't want this to cause v3. It's obviously a bug though.