fix: SuperComponent input filter must use identity, not equality

[Aarkin7]

Proposed Changes:

SuperComponent.run and SuperComponent.run_async filtered out the internal _delegate_default sentinel with value != _delegate_default. That calls __ne__ on every user-provided input, which for numpy arrays, pandas DataFrames/Series, and torch tensors returns an element-wise array, and the dict-comprehension's boolean context then raises ValueError: The truth value of ... is ambiguous. So passing any array-shaped input (a routine ML pipeline thing) failed before the wrapped pipeline ever ran.

Switched both call sites to is not _delegate_default. That's an identity check on a sentinel class — what the filter actually meant, and it never touches __ne__ on user values.

How did you test it?

• Added two unit tests in TestSuperComponentDelegateDefaultFiltering covering the sync (run) and async (run_async) paths with a real numpy.ndarray input.
• hatch run test:unit test/core/super_component/, all 29 tests pass (27 pre-existing + 2 new).
• hatch run test:unit test/core/, 1587 passed, no regressions.
• hatch run test:types haystack/core/super_component/super_component.py: clean.
• hatch run fmt-check: clean.

Notes for the reviewer

• Two-line behavioural change, repeated in run (line 122) and run_async (line 150). I kept a one-line # is not, not != comment above each so the next reader doesn't "simplify" it back.
• Semantic equivalence for ordinary values: _delegate_default is a class object with no __eq__ override, so for any value that doesn't overload __ne__, != and is not produce the same result. Behaviour changes only for (a) array-like values that overload __ne__ to return non-scalar (the fix), and (b) the pathological case of a user class whose __eq__ returns True for everything (previously silently dropped, now correctly kept, strict improvement).
• Release note added under releasenotes/notes/.

Checklist

• I have read the contributors guidelines and the code of conduct.
• I have updated the related issue with new insights and changes.
• I have added unit tests and updated the docstrings.
• I've used one of the conventional commit types for my PR title: fix:, feat:, build:, chore:, ci:, docs:, style:, refactor:, perf:, test: and added ! in case the PR includes breaking changes.
• I have documented my code.
• I have added a release note file, following the contributors guidelines.
• I have run pre-commit hooks and fixed any issue.

[vercel]

@Aarkin7 is attempting to deploy a commit to the deepset Team on Vercel.

A member of the Team first needs to authorize it.

[sjrl]

@bogdankostic for some additional context we did run into a problem like this before in our Pipeline class here #8873

[sjrl]

We have numpy in our test deps so we shouldn't need this and can just import numpy as normal.

[sjrl]

@bogdankostic I can take over the review for this