We intend to measure availability of the (/Authorisation Endpoint) by using an un-registered ADR client , and if the system responds with an error thereby blocking the request to proceed, we will consider that API is functional. This saves us from constructing an ecosystem of test entities (required if we want to proceed to successful issue of token) and it does not add spurious calls to our metrics. We believe it is compliant with the definition of availability on the standards. We would welcome any comment/advice on our intended approach.
The approach you have indicated in your question would be considered compliant by the Data Standards Body.