Flush bad write data from the client socket.

If the client makes a write that's out of range, by the time we get to
validate the message at the server end the client has already stuffed
the socket with data we can't use, so we have to flush it.

This patch also fixes a potential problem in the acceptance tests where
the error field was being returned as an array rather than a value.

tests/acceptance/flexnbd/fake_source.rb

@@ -26,9 +26,17 @@ module FlexNBD
     def read_hello()
       timing_out( FlexNBD::MS_HELLO_TIME_SECS,
                   "Timed out waiting for hello." ) do
-          fail "No hello." unless (hello = @sock.read( 152 )) &&
-            hello.length==152
-        hello
+        fail "No hello." unless (hello = @sock.read( 152 )) &&
+          hello.length==152
+
+        magic_s = hello[0..7]
+        ignore_s= hello[8..15]
+        size_s  = hello[16..23]
+
+        size_h, size_l = size_s.unpack("NN")
+        size = (size_h << 32) + size_l
+
+        return { :magic => magic_s, :size => size }
       end
     end
 
@@ -99,7 +107,7 @@ module FlexNBD
 
       {
         :magic => magic,
-        :error => error_s.unpack("N"),
+        :error => error_s.unpack("N").first,
         :handle => handle
       }
     end