fix(v2): drop rw_timeout ffmpeg option rejected by the RTSP demuxer

Field ffmpeg reported 'Option rw_timeout not found' and exited, causing a
reconnect loop. rw_timeout is an avio/protocol option, not an RTSP demuxer
option; the RTSP socket timeout is build-specific (-stimeout / -timeout).
Omit it for portability — a dead TCP connection still ends decode and
triggers reconnect. Keeps transport and low-latency flags.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

v2/internal/source/stream.go

@@ -278,9 +278,11 @@ func appendInputOptions(args []string, config Config) []string {
 	if transport == "tcp" || transport == "udp" {
 		args = append(args, "-rtsp_transport", transport)
 	}
-	if config.Timeout > 0 {
-		args = append(args, "-rw_timeout", strconv.FormatInt(config.Timeout.Microseconds(), 10))
-	}
+	// NOTE: the RTSP socket-timeout option is build-specific (`-stimeout` on
+	// ffmpeg < 7.0, `-timeout` on >= 5.1) and `-rw_timeout` is rejected by the
+	// RTSP demuxer ("Option rw_timeout not found"). We omit it for portability;
+	// a dead TCP connection still ends the decode and triggers reconnect. A
+	// version-matched timeout can be reintroduced once the ffmpeg build is known.
 	if config.LowLatency {
 		args = append(args, "-fflags", "nobuffer", "-flags", "low_delay")
 	}