Fix test_time test.

[greydot]

Sometimes sleep() may return a bit earlier than expected, again, due to scheduling reasons. This should be accounted for as well.

For example, on my machine I had this:

# **** Solo5 standalone test_time ****
# 
# Overhead of solo5_yield(): 12518 ns
# [5] Sleeping for one second...
# [5] Slept for 999995014 ns
# [5] ERROR: slept too little (expected at least 1000000000 ns)
# [4] Sleeping for one second...
# [4] Slept for 1000012023 ns
# [3] Sleeping for one second...
# [3] Slept for 1000001308 ns
# [2] Sleeping for one second...
# [2] Slept for 1000001248 ns
# [1] Sleeping for one second...
# [1] Slept for 1000001078 ns
# Solo5: solo5_exit(1) called

[dinosaure]

We probably should look deeper and verify that is an expected behavior. I'm not sure that Solo5 should react like this.

[greydot]

Could it depend on hardware, perchance?

[dinosaure]

I will merge this PR but it seems that the initial error hiddens something bigger than what we expect. I will probably adds fews comments and create an issue from this unexpected behavior. Currently, this PR concerns only tests so that's fine to merge it and release it. Thanks!

[reynir]

@greydot do you know on which platform you observed this? Was it spt, hvt, virtio, ...?

[greydot]

@greydot do you know on which platform you observed this? Was it spt, hvt, virtio, ...?

It was spt. Though, I didn't check the others. I'll re-check later.