Add support of applications with distributed memory

[foxtran]

Hello!

As I can see, at the current moment only shared-memory applications are supported (the applications have the same address space). Some applications work on distributed memory (just like client-server model in Tracy) in that the data sharing happens via network or other interfaces. It would be nice to have something that allow users to look on several executables at the same time in the same graph.

At the current time, the client-server diagram looks as follows:

I suppose for enabling this feature, it will be changed to the following:

It will be extremely useful for MPI-applications. Typically, such applications share the same code but different data between MPI threads.

[wolfpld]

See #825 and related issues for discussion on this topic. I've been meaning to take a closer look at this set of PRs, as they need some work, and I don't think PR ping-pong would be appropriate in this case, but sadly other things have been more interesting to me at the (rather long) moment. Getting an extra pair of eyes on how/if this is working properly would be nice.