Insufficient detail logged on connection failure to IdP metadata URL

This issue arose recently when a customer is attempting to register a new SAMP IdP provider.

Nexus was unable to connect to the metadata URL provided, but we do not log enough information to say why the connection failed, returning only "error sending request for url".

Error Response:
status: 400 Bad Request;
headers: {
  "content-type": "application/json",
  "x-request-id": "xxx",
  "content-length": "248",
  "date": "Fri, 12 Nov 2024 22:52:59 GMT"
};
value:Error {
  error_code: Some("InvalidValue"),
  message: "unsupported value for \"url\": error querying url: error sending request for url (https://sso.example- 
  keycloak.com/realms/myrealm/protocol/saml/descriptor)",
  request_id: "xxx"
}

We have a couple similar issues, https://github.com/oxidecomputer/omicron/issues/6717 and https://github.com/oxidecomputer/omicron/issues/6716 around vague connection errors. The underlying issue is probably related to https://github.com/seanmonstar/reqwest/discussions/2342, which has some suggestions around ways to extract more detail.

A systematic way to ensure we're logging sufficient detail on chained errors seems necessary.

Start at nexus/src/app/silo.rs around the linked metadata URL connection code, then compare the related issues 6717 and 6716 and the reqwest discussion 2342. Determine how chained connection failures are currently reported and define completion as returning enough underlying detail to explain why the IdP metadata request failed across these cases.