Gemm Col and Whole Array

[asyms]

Add main files and support to run Gemm workload on 2 tiles of a single core or on 2x2 rows/cols across the array.

Currently, there are still two small issues in the single_col example regarding the number of DMA channels being exceeded. @jorendumoulin please take a look at this when you find the time.

error: "-":4:26: 'aie.tile' op number of output DMA channel exceeded!
 note: "-":4:26: see current operation: %1 = "aie.tile"() <{col = 0 : i32, row = 0 : i32}> : () -> index
error: unknown: 'aie.flow' op attribute 'source_channel' failed to satisfy constraint: 32-bit signless integer attribute whose minimum value is 0
 note: unknown: see current operation: "aie.flow"(%1, %0) <{dest_bundle = 1 : i32, dest_channel = 4 : i32, source_bundle = 1 : i32, source_channel = -1 : i32}> : (index, index) -> ()