more fixes for generic benchmark wrt Corstone-300 (#2592)

@tensorflow/micro

Add person_detection_vela.tflite to tensorflow/lite/micro/models.

Use the ethos-u co-processor and ethos-u enabled model when running benchmark.

bug=fixes #2591

tensorflow/lite/micro/models/person_detect_vela.tflite

diff --git a/tensorflow/lite/micro/models/person_detect_vela.tflite b/tensorflow/lite/micro/models/person_detect_vela.tflite
new file mode 100644
index 0000000..95cf476
--- /dev/null
+++ b/tensorflow/lite/micro/models/person_detect_vela.tflite
Binary files differ

tensorflow/lite/micro/tools/ci_build/test_cortex_m_corstone_300.sh

diff --git a/tensorflow/lite/micro/tools/ci_build/test_cortex_m_corstone_300.sh b/tensorflow/lite/micro/tools/ci_build/test_cortex_m_corstone_300.sh
index 2a62346..ab136e5 100755
--- a/tensorflow/lite/micro/tools/ci_build/test_cortex_m_corstone_300.sh
+++ b/tensorflow/lite/micro/tools/ci_build/test_cortex_m_corstone_300.sh
@@ -44,12 +44,12 @@
 readable_run make -f tensorflow/lite/micro/tools/make/Makefile CO_PROCESSOR=ethos_u OPTIMIZED_KERNEL_DIR=${OPTIMIZED_KERNEL_DIR} TARGET=${TARGET} TARGET_ARCH=${TARGET_ARCH} TOOLCHAIN=${TOOLCHAIN} test
 
 # Run generic benchmark.
-# Instruction cycle counter not functional in simulator for ethos-u co-processor.
 readable_run make -j$(nproc) -f tensorflow/lite/micro/tools/make/Makefile \
+  CO_PROCESSOR=ethos_u \
   OPTIMIZED_KERNEL_DIR=${OPTIMIZED_KERNEL_DIR} \
   TARGET=${TARGET} \
   TARGET_ARCH=${TARGET_ARCH} \
   TOOLCHAIN=${TOOLCHAIN} \
-  GENERIC_BENCHMARK_MODEL_PATH=tensorflow/lite/micro/models/person_detect.tflite \
+  GENERIC_BENCHMARK_MODEL_PATH=tensorflow/lite/micro/models/person_detect_vela.tflite \
   GENERIC_BENCHMARK_ARENA_SIZE=`expr 150 \* 1024` \
   run_tflm_benchmark