drbd: An incapable primary should not outdate a far UpToDate secondary

Let's consider a diskless node A with a resync from C to B. B is
Primary, and C is up-to-date.

   A
  / \
 B<--C

Then the resync is interrupted

   A
  / \
 B   C

B becomes a primary without access to UpToDate data. It either throws
I/O errors or it freezes I/O. In either case, it should not outdate C.

Fix this by detecting B as an incapable primary when evaluating the
data gathered in the empty two-phase commit after losing connection to
a primary.

drbd/drbd_state.c

@@ -5559,8 +5559,9 @@ static bool do_twopc_after_lost_peer(struct change_context *context, enum change
 	struct twopc_reply *reply = &resource->twopc_reply;
 	u64 directly_reachable = directly_connected_nodes(resource, NEW) |
 		NODE_MASK(resource->res_opts.node_id);
+	bool pri_incapable = reply->primary_nodes && !reply->weak_nodes; /* TWOPC_PRI_INCAPABLE */
 
-	if (phase == PH_COMMIT && (reply->primary_nodes & ~directly_reachable)) {
+	if (phase == PH_COMMIT && (reply->primary_nodes & ~directly_reachable && !pri_incapable)) {
 		__outdate_myself(resource);
 	} else {
 		struct drbd_device *device;