COMPMID-452 CL Depthwise Separable Convolution Layer kernel implementation, validation and benchmarking for 3x3xC depthwise filter and DataType::F32.

Change-Id: I95c0c87709763cdbf58d0de66025eac86e30791b
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/82768
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Steven Niu <steven.niu@arm.com>
28 files changed