fix(ws): surface HTTP status/body on relay dial failure

[PrashantBtkl]

Summary

• On an EC2 host, forager logged repeated dial failed: websocket: bad handshake for ~30 minutes during a relay-side connectivity issue, with no way to tell whether the relay was down, rejecting auth, or rate limiting.
• gorilla/websocket returns the HTTP response alongside ErrBadHandshake, but forager discarded it (conn, _, err := ...).
• Now captures and logs the response status code and body (truncated to 1KB) so future incidents show e.g. status=502 vs status=401 vs status=429 instead of an opaque error.

Test plan

• go build ./...
• go vet ./pkg/ws/...
• go test ./pkg/ws/... ./cmd/...

[gemini-code-assist]

Code Review

This pull request enhances error reporting during WebSocket connection failures by capturing and reading the HTTP response body and status code. The reviewer identified a potential nil pointer dereference panic if resp.Body is nil and provided a code suggestion to safely check resp.Body before reading or closing it.

[gemini-code-assist]

To prevent potential nil pointer dereference panics, we should check if resp.Body is nil before attempting to read from or close it. Even if resp.Body is nil, we can still safely report the HTTP status code.

		if resp != nil {
			var body []byte
			if resp.Body != nil {
				body, _ = io.ReadAll(io.LimitReader(resp.Body, 1024))
				resp.Body.Close() // nolint:errcheck
			}
			return fmt.Errorf("dial failed: %w (status=%d body=%q)", err, resp.StatusCode, body)
		}