Fix handling of radio/network overruns (#249)

* Fix handling of radio/network overruns

- The network can get stuck in an available loop if the radio is receiving data faster than the network can process it. If this happens, the RX FIFO needs to be flushed
- Also return NETWORK_OVERRUN when this happens
- Add network system type 160 NETWORK_OVERRUN

* Set timeout to 100 in available()

Author: TMRh20

RF24Network.cpp

@@ -132,14 +132,12 @@ uint8_t ESBNetwork<radio_t>::update(void)
 
     uint8_t returnVal = 0;
 
-    uint32_t timeout = millis();
+    uint32_t timeout = millis() + 100;
 
     while (radio.available()) {
-        if (millis() - timeout > 1000) {
-#if defined FAILURE_HANDLING
-            radio.failureDetected = 1;
-#endif
-            break;
+        if (millis() > timeout) {
+            radio.flush_rx();
+            return NETWORK_OVERRUN;
         }
 #if defined(ENABLE_DYNAMIC_PAYLOADS) && !defined(XMEGA_D3)
         frame_size = radio.getDynamicPayloadSize();

RF24Network.h

@@ -122,6 +122,11 @@
 // NO ACK Response Types
 //#define NETWORK_ACK_REQUEST 192
 
+/**
+ * Messages of this type indicate the network is being overrun with data & the RX FIFO has been flushed.
+ **/
+#define NETWORK_OVERRUN 160
+
 /**
  * Messages of this type signal the sender that a network-wide transmission has been completed.
  *