CI flake: "provided port is already allocated" occassionally happens during e2e cross cluster tests

[AustinAbro321]

We have seen this flake a few times, it is unclear why it is happening.
In all of the cases the error is provided port is already allocated
In the following two cases the error is on the below line.

• https://github.com/defenseunicorns/zarf/actions/runs/8100430794/job/22138566751
• https://github.com/defenseunicorns/zarf/actions/runs/8160962762/job/22308786570

_, initStdErr, err := e2e.Zarf("init", "--components="+initComponents, "--nodeport", "31337", "-l", "trace", "--confirm")
require.NoError(t, err)

We also see a similar error here on here on test 99:

• https://github.com/defenseunicorns/zarf/actions/runs/8012041029/job/21886586280

stdOut, stdErr, err = e2e.Zarf("init", "--components=k3s", "--confirm")
require.NoError(t, err, stdOut, stdErr)

[AustinAbro321]

We believe that during the first two occurrences of the the problem was the registry port 31337 assigned in the test conflicted with the randomly generated trafeik ports created by k3d. This will be solved by disabling trafeik during e2e tests in #2282.

The third issue is an actual bug in zarf. It is caused by the fact that the injector (seed registry) selects a random port and this port could conflict with the registry port. Looking at the logs here confirms this as the seed registry matches the registry in the logs https://github.com/defenseunicorns/zarf/actions/runs/8012041029/job/21886586280.

[lucasrod16]

Hit this again: https://github.com/defenseunicorns/zarf/actions/runs/8841615794/job/24279036176?pr=2456#step:7:494

[AustinAbro321]

Yeah it's only like a 1 in 3000 chance, but it's still worth fixing for the init case. Especially since this can also happen to real uses if they have the proper luck or lack thereof