trace: track suspensions, check they match.

I added this debugging because the next test revealed a mismatch, so
I wanted to see where it was happening.

The comment in lightningd suggests it's possible, but I can't see any
code which suspends in the lightningd io_loop, so I cannot see how
this is triggered.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

lightningd/lightningd.c

@@ -1409,14 +1409,6 @@ int main(int argc, char *argv[])
 	/*~ Now handle sigchld, so we can clean up appropriately. */
 	sigchld_conn = notleak(io_new_conn(ld, sigchld_rfd, sigchld_rfd_in, ld));
 
-	/* This span was started before handing control to `io_loop`
-	 * which suspends active spans in-between processing
-	 * events. Depending on how the `io_loop` was interrupted, the
-	 * current context span may have been suspended. We need to
-	 * manually resume it for this case. Notice that resuming is
-	 * idempotent, and doing so repeatedly is safe.
-	 */
-	trace_span_resume(argv);
 	trace_span_end(argv);
 
 	/*~ Mark ourselves live.