Suggestion: set detector_type to risk_name in response

[declark1]

For Granite Guardian (and presumably Llama Guard), the detector_type field in the detection response is set generically to "risk" rather than the specific risk name that the results are associated with.

As there are default risk types applied for each endpoint, e.g. answer_relevance for /api/v1/text/generation, the risk may be unknown to the user if they didn't request it explicitly as a detector parameter. Setting this field to the specific risk name rather than "risk" would be a better use of this field.