Describe contextual input values propagated the Extensibility Provider (EP) in the design of #3565

Would it be possible to describe what input values (such as the `Authorization` header) are propagated to the Extensibility Provider (EP) ?

In the case of credentials/secrets (like the k8s kubeconfig) what is the trust model?

_Originally posted by @asilverman in https://github.com/Azure/bicep/issues/3565#issuecomment-1106831611_

Start by reading the design discussion in issue #3565 and its linked context. Document which contextual inputs reach the Extensibility Provider, including authorization headers, and explain the trust model for credentials such as a Kubernetes kubeconfig; done means both propagation and credential-handling questions are answered clearly.

Written by the indexing model from the issue text.