Che Operator gets stuck updating endpoint URLs when "legacy" hostname is used

[amisevsk]

Describe the bug

When the Che Operator has to fallback to the legacy format for endpoint hostnames (as added by eclipse-che/che-operator#1702), it can get stuck in a loop updating endpoint hostnames. This occurs because in Golang, map iteration order is random, resulting in a random order number being used for each endpoint (see here.

Che version

7.80@latest

Steps to reproduce

1. Create a workspace with a long metadata.name, so that the default hostname generated for code-redirect-x endpoints is longer than 63 characters (format is <username>-<workspacename>-<endpointname>)
2. Ensure workspace is using legacy format for endpoint hostnames (should instead be <workspaceID>-<number>)
3. Check whether the workspace is able to start.

Expected behavior

We should be using deterministic names for endpoints.

Runtime

Kubernetes (vanilla)

Screenshots

No response

Installation method

other (please specify in additional context)

Environment

other (please specify in additional context)

Eclipse Che Logs

No response

Additional context

Originally reported: #22667
DevWorkspace's workaround for this issue: link

[amisevsk]

This issue was reported as part of #22667 but should have been fixed as of eclipse-che/che-operator#1702 (v7.69) -- cc: @dkwon17 any ideas here? I misunderstood the issue; see updated description.