1.0.7 2026-10-14/17:00

• both: compute rates over the time since the previous run, saved in the
  cache, instead of assuming 300 seconds
• cgroups: read nonvoluntary_ctxt_switches, so involuntary context
  switches are no longer always 0
• cgroups: only total ps stats for included names
• cgroups: merge the IPs and paths of containers in the same podman pod
• cgroups: map docker IDs to names for the cgroupfs driver
• cgroups: find the bridge interface for docker networks
• cgroups: report IPv6 container addresses and look up their gateway
  interface, using ip -j with the args checked before reaching the shell
• cgroups: name non-podman machine.slice units m_$type_$machine instead of
  reading all of machine.slice, which double counted podman containers
• remove the unused ip_to_if and the IO::Interface::Simple dependency
• FreeBSD: report ps output that fails to decode as a error
• oslv_monitor: write the json and snmp files atomically and exit 1 if
  either fails
• oslv_monitor: implement -c and add --backend
• add the lock option and -l, to skip a run while another holds the
  lock instead of clobbering its cache
• oslv_monitor: exit 0 for -h and -v, and 1 for unknown flags
• oslv_monitor: use the OSLV::Monitor version
• require List::Util 1.45 for uniq and drop the unused Math::BigInt
• fix the documented include default
• document new and include, and the internal cgroups subs
• cgroups: usable dies if cgroup v2 is not mounted at /sys/fs/cgroup
• cgroups: always include cache_failure, matching FreeBSD
• FreeBSD: drop the duplicate pid from the ps columns