When running DTS it is assumed that all the required dependencies are already installed and DTS can fail without a very obvious reason. Checking DPDK and Scapy dependencies before setting up may help the tester. These checks could be done as part of the smoke tests, but that'd require to execute them before any setup begins.
Checking without running (I assume you meant this as I think we can't do much more before doing setup, at least with DPDK (I'm thinking of hugepages); I don't think this would be that much of a problem with Python dependencies) may be complicated to do on all systems, how would you propose we do this? Just trying to find the required libraries in the usual places? This could backfire on systems which deviate from that. We'd also need to abstract this between linux/windows, but that's a minor point. I think the only foolproof way to do this would be to try running DPDK and Scapy (or other TGs) which we're already doing in smoke tests (or we wanted to do - if we're missing a basic functional traffic test, we should add it). Testing TG version is an addition thing we could do. If an application fails because of dependencies, the error message from the application should point to that (we can remedy that in DPDK applications). If not, DTS could emit a suggestion to check dependencies.