Different logic for spinning up "server" testlets?

[Mierdin]

Currently, all testlets are started asynchronously, and the number of them that is started is equal to the number of targets (literally len(tr.Targets) in executetestrun.go)

This makes sense for "client-style" testing, but not necessary for servers. Because both server and client testlets are started via the same code in ExecuteTestRunTask, if there are 5 targets, then there will be 5 instances of iperf -s.

Need to think about this some more and answer some questions:

• When ExecuteTestRunTask runs a server testlet, what is tr.Targets? The clients? Or nothing? It's been a while since I wrote that code so I'm wondering if all of this is even true.
• Do we need to come up with some kind of metadata system that tracks what "type" a testlet is? Or is the YAML syntax sufficient? (keeping in mind that some testlets like iperf can do both with different args)
• Do we need a second set of logic for spinning up server type testlets?