[JENKINS-51483] Remove erroneous entry in split-plugin-cycles.txt

[dwnusbaum]

See JENKINS-51483.

This entry should have been removed before #3301 was merged. Here is the jdk-tool:1.0 pom that shows that it does not depend on apache-httpcomponents-client-4-api: https://github.com/jenkinsci/jdk-tool-plugin/blob/jdk-tool-1.0/pom.xml. EDIT: jdk-tool is the wrong place to justify this. The actual justification is that apache-httpcomponents-client-4-api used JDKInstaller in some tests and the entry should never have been added.

Proposed changelog entries

• Bug: Restore implied dependency on JDK Tool Plugin from Apache HttpComponents Client 4 API Plugin

Submitter checklist

• JIRA issue is well described
• Changelog entry appropriate for the audience affected by the change (users or developer, depending on the change). Examples
  * Use the Internal: prefix if the change has no user-visible impact (API, test frameworks, etc.)
• Appropriate autotests or explanation to why this change has no tests
• For dependency updates: links to external changelogs and, if possible, full diffs

Desired reviewers

@reviewbybees

[Wadeck]

🐝 thx for the context / information

[dwnusbaum]

@jglick Would be great if you could confirm whether this will cause any serious issues.

[jglick]

Here is the jdk-tool:1.0 pom that shows that it does not depend on apache-httpcomponents-client-4-api

First of all, the dependency is in the other direction. Second, this file is listing dependencies which are implied by the split list, not defined in the POM, but which are not in fact real—would quietly disappear if the plugin on the left side were simply rebuilt against a newer core version.

[dwnusbaum]

@jglick Thanks for looking!

Second, this file is listing dependencies which are implied by the split list, not defined in the POM, but which are not in fact real—would quietly disappear if the plugin on the left side were simply rebuilt against a newer core version.

In this case, the implied dependency was real (at least when running tests), because apache-httpcomponents-client-4-api had some tests that used JDKInstaller. Now, I realize that the (thankfully removed) dependency on apache-httpcomponents-client-4-api from jdk-tool was totally incorrect because apache-httpcomponents-client-4-api would have needed to depend on jdk-tool for its tests, causing a circular dependency, right? It seems like I shouldn't have added the entry in the first place.

My intention with this PR is to restore the implied dependency. I will update the description and changelog.

[jglick]

To the extent I understand.

[oleg-nenashev]

Test failure is unrelated, INFRA issue

[oleg-nenashev]

👍 since I am the author of those tests which got removed (and this is fine)