EINTR is not crash-worthy

This is how it should have been handled anyway. This fixes a (benign)
race condition when the server is killed, as well as gprof builds.

Also updated tdata.

src/CNProtocol.cpp

@@ -312,6 +312,10 @@ void CNServer::start() {
         // the timeout is to ensure shard timers are ticking
         int n = poll(fds.data(), fds.size(), 50);
         if (SOCKETERROR(n)) {
+#ifndef _WIN32
+            if (errno == EINTR)
+                continue;
+#endif
             std::cout << "[FATAL] poll() returned error" << std::endl;
             terminate(0);
         }

tdata

@@ -1 +1 @@
-Subproject commit e9e03fd0f0ddc09a29db03cc68a5e5276712f2a7
+Subproject commit 9a3991958689087a5846cac71edd16c82edbf09d