Enable sw_emu on Vitis HLS v2021.2

[ueqri]

Summary of making sw_emu work, based on the HLS hardware codes of last release f35c82a:

• Use ap_axiu for AXI-4 stream in kernel interface, and self-defined VEC_AXIS_T for FIFO inside kernel respectively
• Modify stream duplicator & merger to transform between ap_axiu(external stream port) and VEC_AXIS_T(internal stream FIFO)
• Remove exit condition in kernel k2k_relay, to support multiple tests in host program

The implementation is tested on Vitis HLS v2021.2, targeting platform Xilinx Alveo U280.