Update jenkins.init.in

[FIXED JENKINS-20245]

Integrated v2.38

Improved documentation.

[FIXED JENKINS-21394] Avoid irrelevant job queing while node is offline

Conflicts:
	test/src/test/java/hudson/model/ProjectTest.java

Ugly hack to fix destroyProcess for Java8

[FIXED JENKINS-22553] Do not expose plugin list that can be a subject of modification

(It may have a @TestExtension QueueItemAuthenticatorDescriptor if one is needed.)

No sense in spending ~20s running HtmlUnit to verify code which has since been deleted anyway (configuration-time ACL checks).

Test of JENKINS-13502 fix is obsolete since we no longer check permissions when configuring triggers.

or servlet container applies form-login to _all_ URLs 
(detected on tomcat 7.0.50)

(Forgot that Run’s are unmarshalled in place after a special constructor is called, so there is no need for readResolve or other tricks.)
Also calling RunListener.onDeleted outside of the Run lock to avoid problems with things like ChangeLogHistoryRunListener.

[FIXED JENKINS-22570] On a configure screen that has multiple groups of radio buttons, clicking the apply button clears all but the last radio group selection

- Always run git status before and after you commit, just in case there are changes you made since your last call to git add

- At least the integration test cases caught this one though

[FIXED JENKINS-22570] On a configure screen that has multiple groups of radio buttons, clicking the apply button clears all but the last radio group selection

- Changed from mutating the e.name value to maintaining an index and using that index to get the required substring

- Risk of unintended consequences is low. The Submit flow moves off the page before the consequences of the effect
  could be seen... though a careful attention to the page submit can identify the change... and the page is now
  left as before clicking f:apply instead of mutated.

Hoping nobody uses it programmatically.

Also, added @Restricted to the new one.

streams created from SocketChannel doesn't allow concurrent read/write
by different threads, which appear to be a bug in NIO.