Use wall time in throughput test

[jean-roland]

This fixes throughput test using system time instead of wall time, which is what we want.

[milyin]

On the one hand I like this idea: the code becomes more C++ style and uses right tool for the purpose. On the other hand this example matches equal example in zenoh-c (https://github.com/eclipse-zenoh/zenoh-c/blob/master/examples/z_sub_thr.c), so it's better to use exactly the same measuring method. On the third hand :-) the corresponding Rust test works in the rust's way to measure the time, so why not to do in here in C++ way.

Anyway it's better to 1) make sure that measuring results are the same and 2) integrate it after the release

[jean-roland]

We deviate from zenoh-c but we're now closer to the zenoh-cpp own zenoh-ping example. Anyway, I verified that the measurements were correct and I'm ok with integrating this after the release.