[Dataflow][Examples] Provide a convolution example for FEATHER

[Oh-Afterglow]

Description

This pull request provides a FEATHER example of performing a 2D convolution operation and switching the layout from channel-last to row major at the same time, which is illustrated in Fig.11 of the paper.

The FEATHER kernel in this example is mostly the same as the GEMM example, except for the parameters determining the sizes of input buffers.

The configurable parameters of this example includes NEST parameters AW and AH, and the 7 dimensions (NCHWMRS) of convolution operation. The source code contains more detailed explanation of the parameters.

Checklist

Please make sure to review and check all of these items:

• PR's title starts with a category (e.g. [Bugfix], [IR], [Builder], etc)
• All changes have test coverage (It would be good to provide ~2 different test cases to test the robustness of your code)
• Pass the formatting check locally
• Code is well-documented

[chhzh123]

In this case, we should put gemm and conv in the same file?

[Oh-Afterglow]

I also think so, although the GEMM example needs some modifications before merging, since its number of PE local registers (arbitrarily configurable) is not exactly complied with the original FEATHER design (=AH). The FEATHER implementation in this convolution example is a more correct one.