Add a system call return test that tries all syscall types

[bradjc]

Matches tock/tock#5083

To accomplish this, this PR also:

• Adds the helper functions for all of the remaining command return types.
• Implements the converters for rv64 support.
• Fixes the riscv command implementation which was using only u32 types.
• Adds a test which calls each command/subscribe/allow and checks the return values.

[brghena]

One question for you right now

[brghena]

I'm trying to decide if these failure variants are generally useful or only applicable to your test application.

Let's say I was building a capsule which either returns success with U32 and U64 or failure with U64. Is this how you'd propose it be implemented? Any problems with this?

syscall_return_t cval = command(...);

returncode_t ret = tock_command_return_u32_u64_to_returncode(cval, u32_ptr, u64_ptr);
if (ret != RETURNCODE_SUCCESS) {
  return tock_command_return_failure_u64_to_returncode(cval, u64_ptr);
}

For IPC, I made a tock_command_return_u32_u64_or_u64_to_returncode() function, but I'll admit that it did feel quite unwieldy. I could replace it with the above snippet.

[bradjc]

Oh, interesting, you know I hadn't actually given this any thought. It seems wrong that we wrote the original converters assuming a default failure-no-value alternative without including that in the function name.

I kind of think what you did for IPC is what this all should look like. It's a little unwieldy, I guess, but it is what we want to express from syscalls: they have a single return success/failure variant.

But, I'm not sure how important it is to resolve this right now.

[brghena]

Well, the reason I asked was because I was trying to figure out what the purpose of the Failure variants are at all. I think they're useful for testing, for what I described above, and for nothing else.