iter29 DIAG: dump trailing 80 bytes of HEVC slice_data per slice

Env-gated via LIBVA_HEVC_DUMP_SLICE_TAIL=1. Goal: characterise the
40-byte inflation in libva's slice_data buffer vs ffmpeg-v4l2request
(see iter27/28 close — HEVC frame 2+ divergence at byte 1382401).

Dumps per slice: nal_unit_type, slice_data_size, slice_data_byte_offset,
and the last 80 bytes of source_data for that slice. Lets us see if the
trailing 40 bytes are (a) real entropy, (b) trailing zeros, (c) a
next-NAL start code prefix, or (d) random memory.

src/h265.c

@@ -70,6 +70,7 @@
 #include "surface.h"
 
 #include <assert.h>
+#include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
@@ -613,6 +614,26 @@ int h265_set_controls(struct request_data *driver_data,
 				       cumulative_offset,
 				       &slice_params_array[i]);
 
+		/* iter29 DIAG: dump trailing 80 bytes of each HEVC slice.
+		 * Set LIBVA_HEVC_DUMP_SLICE_TAIL=1 to enable. Goal: characterise
+		 * the 40-byte inflation in ffmpeg-vaapi vs ffmpeg-v4l2request for
+		 * frame 2+ slices (see iter27/28 close). */
+		if (getenv("LIBVA_HEVC_DUMP_SLICE_TAIL")) {
+			uint32_t sz = slice->slice_data_size;
+			uint32_t boff = slice->slice_data_byte_offset;
+			uint8_t *p = (uint8_t *)surface_object->source_data + cumulative_offset;
+			uint32_t dump_n = sz < 80 ? sz : 80;
+			uint32_t start = sz - dump_n;
+			uint32_t k;
+			fprintf(stderr, "iter29 slice[%u] nut=%u size=%u boff=%u start_in_slice=%u tail80:",
+				i, slice_params_array[i].nal_unit_type, sz, boff, start);
+			for (k = 0; k < dump_n; k++) {
+				if ((k & 0xf) == 0) fprintf(stderr, "\n  +%04x:", start + k);
+				fprintf(stderr, " %02x", p[start + k]);
+			}
+			fprintf(stderr, "\n");
+		}
+
 		cumulative_offset += slice->slice_data_size;
 	}