`_SuperComponent` always define `run_async` even when the underlying Pipeline is sync

[anakin87]

Discovered in https://github.com/deepset-ai/haystack/pull/9420#pullrequestreview-2864343991

from haystack.components.preprocessors import DocumentPreprocessor
from haystack import Document
import asyncio

preprocessor = DocumentPreprocessor()

preprocessor.warm_up()

asyncio.run(preprocessor.run_async(documents=[Document(content="something")]))

Raises: TypeError: Pipeline is not an AsyncPipeline. run_async is not supported.

---

_SuperComponent defines a run_async method even when the underlying pipeline is synchronous. This leads to runtime errors and potentially confuses users, as the method appears available but always fails.

[sjrl]

@anakin87 I see a few ways to solve this:

• Auto convert a Pipeline to AsyncPipeline under-the-hood if a user calls run_async on a SuperComponent defined with Pipeline. And I guess vice-versa if the opposite happens.
• Consolidate our Pipeline and AsyncPipeline abstractions into one object? Not sure how easy this would be.
• Dynamically add the correct run method through the decorator? This approach wouldn't work with the inheritance approach creating generic SuperComponents when using SuperComponent directly.
• Leave as-is but throw an error at run time if a user tries to use run_async but defined their SuperComponent with Pipeline

What do you think?