Fix incorrect retry when SIGINT reaches ssh

Make sure that the retry loop is never activated if a SIGINT has been
received. In some cases the SIGINT could have been incorrectly detected
as a lost connection which would normally enable the retry loop. This
can happen when the ssh process receives SIGINT as part of the process
group and shuts down, which to us looks like a lost network connection.

src/remote.ml

@@ -1880,6 +1880,9 @@ let buildShellConnection shell host userOpt portOpt rootName termInteract =
 
            Don't let these signals reach ssh by blocking them.
 
+           Unfortunately, a bug introduced in OpenSSH 9.6 (also present in 9.7)
+           breaks this workaround by unblocking SIGINT in the ssh process.
+
            The signals could be ignored instead of being blocked because ssh
            does not set handlers for SIGINT and SIGQUIT if they've been ignored
            at startup. But this triggers an error in ssh. The interactive

src/uitext.ml

@@ -786,6 +786,9 @@ let verifyMerge title text =
       true
   end
 
+let intrcount = ref 0
+let intrRequested () = !intrcount <> 0
+
 type stateItem =
   { mutable ri : reconItem;
     mutable bytesTransferred : Uutil.Filesize.t;
@@ -866,7 +869,6 @@ let doTransport reconItemList numskip isSkip =
   in
   Uutil.setProgressPrinter showProgress;
 
-  let intrcount = ref 0 in
   let sigtermHandler _ =
     if !intrcount >= 3 then raise Sys.Break;
     Abort.all ();
@@ -1702,7 +1704,7 @@ and start2 () =
     exit exitStatus
   with
   | Sys.Break -> terminate ()
-  | e when noRepeat || breakRepeat e -> begin
+  | e when noRepeat || breakRepeat e || intrRequested () -> begin
       handleException e;
       exit Uicommon.fatalExit
     end