ConditionalRouter silently coerces string output to another type when output_type is str | None #12617

Describe the bug
ConditionalRouter.run() skips ast.literal_eval coercion when output_type is exactly str, but not when output_type is a Union that includes str (e.g. str | None, Optional[str]). This means a perfectly normal "optional string output" declaration silently has its string value mangled if the rendered value happens to parse as a Python literal.

Affected code
haystack/components/routers/conditional_router.py, inside run():

with contextlib.suppress(Exception):
    if not self._unsafe and output_type is not str:
        output_value = ast.literal_eval(output_value)

This check only special-cases the exact str type. Any Union including str still evaluates output_type is not str as True.

To Reproduce

from haystack.components.routers import ConditionalRouter

routes = [
    {
        "condition": "{{True}}",
        "output": "{{value}}",
        "output_name": "result",
        "output_type": str | None,
    }
]
router = ConditionalRouter(routes, validate_output_type=True)
result = router.run(value="42")
# Expected: {"result": "42"}
# Actual: raises ValueError (int 42 doesn't match str | None)

Without validate_output_type=True, this silently returns 42 (int) instead of "42" (str) to the next pipeline component.

Context
PR #12322 fixed this same class of bug for the exact output_type=str case (see test_string_output_type_preserved_over_literal_eval), but didn't generalize to Unions that include str, which is a very normal way to declare an optional string output.

Proposed fix
Add a helper that checks whether str is output_type itself or a member of a Union output_type, and use that instead of the direct is not str check. I have a fix and passing regression test ready — happy to open a PR.

Environment
Haystack main branch, Python 3.12.4

👋 Hello there! This issue will be handled internally and isn't open for external contributions. If you'd like to contribute, please take a look at issues labeled contributions welcome or good first issue. We'd really appreciate it!