Rethrow abort exception on IoContext::run

We're seeing errors getting reported from actors trying to run promises
without an active request. We've determined this is from tasks trying to
run after an actor's IO context has been aborted. This change stores the
abort exception in the `IoContext` and rethrows it in `IoContext::run`.

Author: ObsidianMinor

src/workerd/io/io-context.h

@@ -336,6 +336,7 @@ class IoContext final: public kj::Refcounted, private kj::TaskSet::ErrorHandler
 
   // Force context abort now.
   void abort(kj::Exception&& e) {
+    abortException = kj::cp(e);
     abortFulfiller->reject(kj::mv(e));
   }
 
@@ -852,6 +853,7 @@ class IoContext final: public kj::Refcounted, private kj::TaskSet::ErrorHandler
 
   DeleteQueuePtr deleteQueue;
 
+  kj::Maybe<kj::Exception> abortException;
   kj::Own<kj::PromiseFulfiller<void>> abortFulfiller;
   kj::ForkedPromise<void> abortPromise = nullptr;
 
@@ -1013,6 +1015,12 @@ kj::PromiseForResult<Func, Worker::Lock&> IoContext::run(
 template <typename Func>
 kj::PromiseForResult<Func, Worker::Lock&> IoContext::run(
     Func&& func, kj::Maybe<InputGate::Lock> inputLock) {
+  // Before we try running anything, let's make sure our IoContext hasn't been aborted. If it has
+  // been aborted, there's likely not an active request so later operations will fail anyway.
+  KJ_IF_SOME(ex, abortException) {
+    kj::throwFatalException(kj::cp(ex));
+  }
+
   kj::Promise<Worker::AsyncLock> asyncLockPromise = nullptr;
   KJ_IF_SOME(a, actor) {
     if (inputLock == kj::none) {