feat: print IPAM resources as tables that say what they hold

[scotwells]

Problem

datumctl get ippools prints a column of generated names beside a column of RFC3339 timestamps, and nothing else:

NAME                                                              CREATED AT
datum-subnet-ipv6-datum-cloud-us-central-1-global-mesh-282a69ab   2026-09-22T20:25:15Z

Every kind used rest.NewDefaultTableConvertor, whose output is hardcoded to the name and the creation timestamp. You cannot tell what range a pool holds, which family it is, or how full it is without a second command per row.

Change

Each kind defines its own columns, mirroring what the milo-ipam plugin already prints for the same resources so the two surfaces agree. Age uses the elapsed-time form every other Kubernetes resource uses; the absolute time is still in -o yaml.

NAME                                                              CIDR                  FAMILY   CLASSES             UTILIZATION   AGE
datum-vpc-ipv6-root                                               2001:db8::/32         IPv6     datum-vpc-ipv6      <0.1%         4d
datum-subnet-ipv6-datum-cloud-us-central-1-global-mesh-282a69ab   2001:db8:a08:1::/64   IPv6     <none>              100.0%        4h20m

Fields that answer a follow-up rather than identify the object — parent, provisioning class, scope, phase — are marked wide, so they appear under -o wide and not before. Utilization distinguishes 0% from <0.1%: a /64 handed out of a /32 rounds to zero but the pool is not untouched.

This fixes every client at once — kubectl, datumctl, the portal — because they all ask the server for the table.

Left out

No API types change, so no regeneration. Worth noting this was invisible until v0.4.1: while every provisioned pool read 0001-01-01T00:00:00Z, the column was uniformly useless and the missing ones did not stand out.