[bgp-config] fix bgp_config_get by name

[nicolaskagami]

Adds .filter(bgp_config::time_deleted.is_null()) to bgp_config_get, with some minor refactoring for readability. Also adds a regression test.

[nicolaskagami]

While writing a unit test for the #10566, I noticed that bgp_config_get was failing to find the config by name, which led me to find #6471.

Apparently we don't expose GET:

Usage: oxide system networking bgp config [OPTIONS] <COMMAND>

Commands:
  create  Create BGP configuration
  delete  Delete BGP configuration
  list    List BGP configurations
  help    Print this message or the help of the given subcommand(s)

However, if we were to expose it or try to get by name internally we'd bump into the same problem as delete had with the full scans.

This change is fine unless we want GET to return entries with non-null time_deleted for some reason.

I also took the liberty of refactoring it a little.

[jgallagher]

LGTM, just a couple suggestions on a little more test fidelity.

[jgallagher]

Could we add assertions that the properties of the config we get back match what we expect based on the create?

[nicolaskagami]

Sure. This would expand the test from a regression test into also checking GET functionality.

[jgallagher]

Could we add another bgp_config_get() after this delete and confirm we get the expected error (presumably a "not found")?

[nicolaskagami]

Sure. This would expand the test from a regression test into also checking GET functionality.

Same here but for DELETE