dmamover 1.0

dma segmentation is moved to dma_mover.
Added tri_benchmark_control (not tested, because tri doesn't work)
Built tcp_cmac and verified that send/recv works on XACC. Throughput is lower w.r.t. MCU based collectives.
Built tri design but it doesn't work

• test self send/recv:
  • 0->0 : on 1st pkt it works
  • next pkts it fails and spare buffer goes in to an error state (the dma_dequeue receives an error)
  • 1 - >1: pkts goes to 0.
  • Need to put ILAs on AXIS switch and on pkts I suspect something is wrong with tdest there