Running sudo make test-interfaces intermittently failed

dhclient/dhcp6c PID race after VRF move, process_named_running() returns a PID the instant it sees a match, with no stability check. When an interface moves back to the default VRF, dhclient/dhcp6c restarts (T5103) and can exit/respawn again before the test reads /proc/<pid>/cmdline, raising FileNotFoundError instead of a normal assertion. Fixed by re-resolving the PID on failure instead of trusting the stale one.

FRR commit retry swallows errors, commit_configuration() retries frr-reload up to 5 times with zero backoff to work around known FRR races (frr#10132/#10133), but on exhausting retries raised ConfigurationNotValid, which no conf-mode script catches, so it escaped uncaught, surfacing only as an opaque [[vrf]] failed.

Unbounded commit-lock wait in base_vyostest_shim.py, utils/commit.py; the smoketest harness polled the commit lock with no timeout, so a lock left held by #2's failure silently hung an unrelated later test for 20 minutes. Bounded the test harness's wait to 60s