[ring-jetty-adapter] No control over exceptions thrown when obtaining item from ISeq response body

Consider the following code returning an ISeq body that throws after some iterations:

(defn now-millis [] (str "data:" (System/currentTimeMillis) "\n\n"))

(def counter (atom 0))

(defn now-throw []
  (swap! counter inc)
  (if (< @counter 5000) "" (throw (ex-info "Test exception" {}))))

(def body (->> (repeat now-millis)
               (interpose now-throw)
               (map (fn [f] (f)))
               lazy-seq))

(defn handler [request]
  {:status 200
   :headers {"Content-Type" "text/event-stream"
             "Cache-Control" "no-cache"
             "Connection" "keep-alive"}
   :body body})

When I use Curl (acceptable result):

$ curl localhost:8082

data:1751448129717

data:1751448129717
...
...
curl: (18) transfer closed with outstanding read data remaining

At the server end, I see this exception logged (acceptable result):

[qtp1107779742-52] WARN org.eclipse.jetty.ee9.nested.HttpChannel - /
clojure.lang.ExceptionInfo: Test exception {}
    <stack-trace>

The issue is: I do not have control over how to perform cleanup actions when such an exception is thrown.

Start in the ring-jetty-adapter path that handles an ISeq response body, using the provided handler and curl reproduction to observe the exception. Trace how the Jetty-side failure is surfaced and determine where cleanup control could be exposed. Done means callers can perform cleanup when iteration throws, with behavior covered by a focused reproduction or test.