[Ops] Support for downsample op (ttnn.downsample)

[kamalrajkannan78]

• Downsample op exists in ttnn -> https://docs.tenstorrent.com/tt-metal/latest/ttnn/ttnn/api/ttnn.downsample.html#ttnn.downsample. But support for single channel input tensor is not yet done in ttnn -> [Bug Report] ttnn.downsample op - Support Needed for Single Channel Input Tensors (input_channels=1) tt-metal#15563
• Downsample op should be supported in MLIR for single channel input tensor with scale factor 4 and 2. More specifically Input size size is [1,1,480,640] & output sizes should be are [1,1,120,160] , [1,1,240,320]