Don't reset retirement_requester after end of retirement

[d-m-u]

Because we have a column name and a method name that are equivalent, finish_retirement used to reset the retirement_requester to nil. Changing the method to call a different method that only sets the field name prevents this from being reset to nil at the end.

Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1638502

[d-m-u]

@miq-bot add_label bug, blocker
@miq-bot assign @gmcculloug
@miq-bot add_reviewer @tinaafitz
@miq-bot add_reviewer @mkanoor

[d-m-u]

@miq-bot add_label hammer/yes

[gmcculloug]

Looks good to me.

The one thing I would suggest, in a followup PR, is to update the #mark_retired tests to use a Timecop.freeze block around the retires_on date check instead of testing against a large time range.

For example, the check
expect(@service.retires_on).to be_between(Time.zone.now - 1.hour, Time.zone.now + 1.second)

Should really be an exact match if time is frozen.

This currently happens in all three specs being updated here.

[tinaafitz]

@d-m-u Looks good.

[simaishi]

Hammer backport details:

$ git log -1
commit bbf21fc9ccaf0ed3d4d9609b7703ee4b662989ad
Author: Greg McCullough <gmccullo@redhat.com>
Date:   Thu Jan 3 13:52:37 2019 -0500

    Merge pull request #18325 from d-m-u/fixing_retirement_requester_reset
    
    Don't reset retirement_requester after end of retirement
    
    (cherry picked from commit 2547ccaa7c9d0284ae1127585548f3a060f88ccb)
    
    Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1638502

[d-m-u]

Can we get this backported to g for https://bugzilla.redhat.com/show_bug.cgi?id=1641812, please?
@miq-bot add_label gaprindashvili/yes

[d-m-u]

🦃 🐌 🐍 🗡

[simaishi]

@d-m-u Tried backporting to gaprindashvili branch, but entire file is show up as conflicting for spec/models/orchestration_stack/retirement_management_spec.rb Can you please create a separate PR?

[d-m-u]

@simaishi Yes ma'am, just a minute.

[simaishi]

Backported to Gaprindashvili via #18376