fix: reset error on http-request retry

[simonostendorf]

Problem

When an API request fails with a 401 Unauthorized and the client successfully refreshes the token and retries the request, the retried request may return a 200 OK with valid data, but the error is still returned. This is because the error fields (ErrorStatus, Error, ErrorMessage) are populated from the first failed response and are never cleared before the retry.

The following log output illustrates the issue. The first request fails with a 401, the token is refreshed, the retry succeeds with a 200 and valid data but the user still receives status=401 error=Unauthorized message=Invalid OAuth token:

helix: request GET https://api.twitch.tv/helix/eventsub/subscriptions
helix: response 401 {"error":"Unauthorized","status":401,"message":"Invalid OAuth token"}
helix: request POST https://id.twitch.tv/oauth2/token?...&grant_type=client_credentials&scope=...
helix: response 200 {"access_token":"...","expires_in":4985140,...}
helix: request GET https://api.twitch.tv/helix/eventsub/subscriptions
helix: response 200 {"total":1,"data":[...],...}
// App access token refreshed callback fires
ERROR: failed to get event sub subscriptions: status=401 error=Unauthorized message=Invalid OAuth token

Root Cause

In doRequest, when a 401 response is received, the error body is deserialized into resp, setting resp.ErrorStatus, resp.Error, and resp.ErrorMessage. The client then refreshes the token and continues the loop. On the next iteration the request succeeds and resp.Data is populated correctly, but nothing resets the error fields from the previous iteration. The Response struct ends up in a mixed state: valid data alongside a stale 401 error.