Implemented full e2e testing

Test/Integration/TestCommon.h

@@ -1,6 +1,7 @@
 #ifndef TESTCOMMON_H
 #define TESTCOMMON_H
 
+#include "BasicUDPSocket.h"
 #include "CompilerTypes.h"
 #include "StreamString.h"
 
@@ -8,6 +9,48 @@ namespace MARTe {
     extern const char8 * const debug_test_config;
     bool SendCommandGAM(uint16 port, const char8* cmd, StreamString &reply);
     void TestMessageCommand();
+
+    /**
+     * @brief Reads DebugService trace telemetry off @p sock until a real
+     * UDPS DATA packet (Common/UDP/UDPSProtocol.h) is decoded or
+     * @p overallTimeoutMs elapses, and extracts the value of the first
+     * (and, for these single-signal integration tests, only) traced
+     * signal — a scalar at wire offset 0, right after the 8-byte HRT
+     * timestamp that opens every DATA payload.
+     *
+     * DebugService's real UDP telemetry (DebugService::Streamer) speaks the
+     * shared UDPS protocol, not the legacy TraceHeader format declared in
+     * DebugCore.h (that struct is now dead code, kept only for historical
+     * reference) — callers must not reinterpret raw datagrams as
+     * TraceHeader. CONFIG packets (sent once when the traced-signal set
+     * changes) are silently skipped so callers don't need to special-case
+     * them.
+     *
+     * @param sock DebugService's StreamPort/UdpPort listener, already Open()+Listen()ed.
+     * @param overallTimeoutMs total time budget across all read attempts.
+     * @param value output: the decoded scalar value.
+     * @return true if a DATA packet was decoded within the budget.
+     */
+    bool ReadUDPSTraceScalar(BasicUDPSocket &sock, uint32 overallTimeoutMs, uint32 &value);
+
+    /**
+     * @brief Grows the OS receive buffer (SO_RCVBUF) on an already-Open()ed
+     * UDP socket.
+     *
+     * BasicUDPSocket never touches SO_RCVBUF, so listeners are left at the
+     * Linux default (net.core.rmem_default, ~208 KiB on most systems). At
+     * sustained 1 kHz+ trace rates a single scheduling delay in the test
+     * process (e.g. a printf, a GC pass, or just being descheduled) is
+     * enough to overflow that default buffer, causing silent kernel-level
+     * datagram drops (visible as RcvbufErrors in /proc/net/snmp) that show
+     * up as false "discontinuities" even though DebugService sent every
+     * sample. UDPSClient already works around this for production code
+     * (UDPSClient::SetRecvBufferSize); integration tests that Listen()
+     * directly on a raw BasicUDPSocket need the same treatment.
+     *
+     * Best-effort: a failure here just leaves the OS default in place.
+     */
+    void GrowUDPRecvBuffer(BasicUDPSocket &sock, uint32 bytes = 4194304u);
 }
 
 #endif