[teleport-update] status subcommand

[sclevine]

This PR implements the teleport-update status subcommand. Running this command outputs a combination of local and remote status fields that describe the state of agent auto-updates from this agent's perspective.

Additionally, this PR moves all config management logic into config.go. The previous contents of config.go are now in setup.go.

Depends on: #49307

---

The teleport-update binary will be used to enable, disable, and trigger automatic Teleport agent updates. The new auto-updates system manages a local installation of the cluster-specified version of Teleport stored in /var/lib/teleport/versions.

RFD: #47126
Goal (internal): https://github.com/gravitational/cloud/issues/10289

Example:

ubuntu@legendary-mite:~$ ./teleport-update status
proxy: teleport.example.com
enabled: true
pinned: true
active_version: 16.4.3
backup_version: 16.4.1
target_version: 17.0.1
flags:
    - Enterprise
in_window: true
jitter: 1m0s

[zmb3]

Consider trace.ConvertSystemErr for filesystem operations.

[sclevine]

It seems like that would convert the os/fs errors to trace-specific types that this package isn't looking for? Or is the primary intent to avoid stack traces for these errors?

I haven't found any guidelines on when trace should be used or avoided. I try to avoid everything besides trace.Errorf and trace.Wrap (unless I'm writing code that is called by something that expects a specific trace error). That said, happy to follow repo conventions.