Extends TestProxyRulesInUserNS with a new TestProxyTopologyInUserNS test
that builds the ns-pod1 <-> ns-kproxy <-> ns-pod2 topology described in
issue 130926 using veth pairs inside an unprivileged user+network namespace.
The topology uses two veth pairs:
veth0 (10.0.1.1/24 in ns-kproxy) <-> veth0-peer (10.0.1.2/24 in ns-pod1)
veth1 (10.0.2.1/24 in ns-kproxy) <-> veth1-peer (10.0.2.2/24 in ns-pod2)
This is phase 2 of the rootless kube-proxy integration testing work.
Phase 3 will install real Service/Endpoint rules via the proxier and
verify packet forwarding behaviour end-to-end.
Related to kubernetes/kubernetes#130926
Add RunInUserNS() helper to pkg/proxy/util/testing that re-executes
the current test binary inside an unprivileged user+network namespace,
mapping the caller to uid/gid 0 inside the namespace.
Use the helper in a new TestProxyRulesInUserNS test that verifies
nftables rule installation succeeds inside the namespace without
requiring the test runner to be root on the host.
This is the first step toward the topology described in issue 130926:
ns-pod1 <-> ns-kproxy <-> ns-pod2
Subsequent PRs will add veth pairs between namespaces and test
real packet forwarding behaviour end-to-end.
Kartavya Sonar