Integrate improved CPU depthwise convolution kernels

* Replace assembly kernels for depthwise convolution with more optimized
  ones.
* Add int8 assembly kernels.
* Fix implicit padding on optimized kernels

Resolves: COMPMID-3867, COMPMID-4361

Change-Id: I0b0867e05f61be4f368f62190d55e14d0ab3ebf2
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5622
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
diff --git a/arm_compute/core/utils/quantization/AsymmHelpers.h b/arm_compute/core/utils/quantization/AsymmHelpers.h
index cbf7559..c9d0930 100644
--- a/arm_compute/core/utils/quantization/AsymmHelpers.h
+++ b/arm_compute/core/utils/quantization/AsymmHelpers.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017-2020 Arm Limited.
+ * Copyright (c) 2017-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -89,7 +89,6 @@
  * @param[in]  input                  Input tensor info.
  * @param[in]  weights                Weights tensor info.
  * @param[in]  output                 Output tensor info.
- * @param[in]  idx_ofms               Dimension index to get OFMs from the weights tensor.
  * @param[out] output_multipliers_ptr Pointer to the buffer where to store per-channel multipliers.
  * @param[out] output_shifts_ptr      Pointer to the buffer where to store per-channel shifts.
  *
@@ -98,7 +97,6 @@
 void compute_quantized_multipliers_and_shifts(const ITensorInfo *input,
                                               const ITensorInfo *weights,
                                               const ITensorInfo *output,
-                                              unsigned int       idx_ofms,
                                               int32_t           *output_multipliers_ptr,
                                               int32_t           *output_shifts_ptr);