NXP backend: Add quantization of fused biassless convolution to integration tests pipeline

Author: StrycekSimon

backends/nxp/tests_models/utils.py

@@ -33,6 +33,9 @@
 )
 from executorch.backends.nxp.quantizer.neutron_quantizer import NeutronQuantizer
 from executorch.backends.nxp.tests_models.model_input_spec import ModelInputSpec
+from executorch.backends.transforms.quantize_fused_convbn_bias_pass import (
+    QuantizeFusedConvBnBiasAtenPass,
+)
 from executorch.devtools.visualization.visualization_utils import (
     visualize_with_clusters,
 )
@@ -149,6 +152,11 @@ def to_quantized_edge_program(
 
     exir_program_aten_quant = convert_pt2e(m)
 
+    if use_qat:
+        exir_program_aten_quant = QuantizeFusedConvBnBiasAtenPass(
+            default_zero_bias=True
+        )(exir_program_aten_quant).graph_module
+
     # To ATen
     core_aten_ep = _to_core_aten(
         exir_program_aten_quant, example_input, None, verbose=True