testing: re-implement some carp tests in python

[melifaro]

Summary

This change is a side-effect of debugging some carp behaviour.
Effectively it's a rewrite of carp:carp_basic_v[46] tests.
I'm not insisting to commit this. However, while it's doesn't provide much more value that the shell version, it can serve as a base for implementing more complex carp tests if so desired. Happy to discuss :-)

[melifaro]

apply black.

[kp]

tests/sys/netinet/carp.py

105

Should we have some magic in TOPOLOGY handling so 'type: bridge' can just add interfaces to the bridge for us?

[melifaro]

tests/sys/netinet/carp.py

105

Sounds interesting. That's not hard to implement.

User can the setup topology as

TOPOLOGY = {
    "vnet1": {"ifaces": ["epair2", "epair3", "bridge"]},
    "vnet2": {"ifaces": ["epair2"]},
    "vnet3": {"ifaces": ["epair3"]},
    "epair2": {"prefixes4": [("", "192.0.2.202/29")]},
    "epair3": {"prefixes4": [("", "192.0.2.203/29")]},
    "bridge": {"type": "bridge", "members": ["epair2", "epair3"], "prefixes4": [("192.0.2.4/29", "")]},
}

I'll look into that

[kp]

To address the actual point of this review: I'm ambivalent. I still think Python doesn't contribute much for these specific tests (although there are cases where it does make more sense, such as the test_source_mac() test here). It also doesn't really seem significantly worse than the shell version, hence the ambivalence.