kernel/riscv64:Added support for omatcopy on RISCV64_ZVL256B

[guoyuanplct]

I modified the omatcopy implementation for riscv64_zvl256b, achieving a certain performance improvement.

[guoyuanplct]

It might be worth considering moving this test — which appears to occasionally hang — to the end, or perhaps adding a timeout (e.g., 3 minutes) to allow for a safe restart if necessary.

[guoyuanplct]

I tried a function that restarts the deadlock test after it gets hanged. This function restarts the program if it runs beyond a certain time limit, with a maximum of 10 retries. In my forked branch, I did some basic testing. On average, the number of retries was around 5 to 8 times. Or we can continue to increase the number of restarts, since each run is very short and won't have a significant impact on the overall time.

[martin-frbg]

Okay, thanks, let's go with your version then - certainly more elegant than my quick fix of letting the entire job time out faster. Eventually this can hopefully be fixed by a newer qemu, or changes to the virtual hardware environment of the job.
(The hanging utest is most likely the fork-and-run-dgemm one, which I haven't seen hang on actual hardware)