fix: correct Bearer token formatting in UMA rego policy for Keycloak

[silvi-t]

Summary

• Fix Bearer token Authorization header formatting in UMA OPA rego policy to comply with stricter parsing in Keycloak 26.5.4+ (CVE-2026-0707)

Details

Keycloak 26.5.4 includes a fix for CVE-2026-0707 ("Incorrect Behavior Order: Authorization Before Parsing and Canonicalization"), which enforces stricter Authorization header parsing per RFC 6750.

The OPA rego policy in test_authorization_services.py was constructing malformed Authorization headers using concat(" ", ["Bearer ", token]). Since OPA's concat joins array elements with the separator, and "Bearer " already has a trailing space, this produced "Bearer <token>" (double space).
The old Keycloak parser was lenient and accepted this; the new parser rejects it.

Verification steps

poetry run pytest -vv testsuite/tests/singlecluster/authorino/authorization/opa/test_authorization_services.py
poetry run pytest -vv --standalone testsuite/tests/singlecluster/authorino/authorization/opa/test_authorization_services.py