fix(multi-networks): configure scylla correctly with the right interface

[fruch]

• configure listen_interface in scylla.yaml as needed
• return correct selected interface on get_nic_devices()
• run scylla_sysconfig_setup again on right network interface

Testing

• 🟢 https://argus.scylladb.com/tests/scylla-cluster-tests/13d394e8-b293-449c-85ee-876322dcfcbd

PR pre-checks (self review)

• I added the relevant backport labels
• I didn't leave commented-out/debugging code

Reminders

• Add New configuration option and document them (in sdcm/sct_config.py)
• Add unit tests to cover my changes (under unit-test/ folder)
• Update the Readme/doc folder relevant to this change (if needed)

[juliayakovlev]

What is it wrong, why this change is needed?

[fruch]

What is it wrong, why this change is needed?

while investigating the job using it, I've seen that we configure all of the CPU pinning in regards of eth0
I thought it might the cause of some other issues as well I've seen in that case, but so far I showed that most of them aren't related to this change.

but it's the correct thing todo, to make sure scylla is configured correctly.