Split expanded_environment_name away from variable expansion

The following discussion from !202162 (merged) should be addressed:

• @fabiopitino started a discussion:

  Idea for a follow-up:

  The problem with expanded_environment_name method is that it has more than 1 responsibility:

  1. It returns the name of the persisted environment.
  2. It expands the user-provided environment name using the job variables.

  I think that these 2 responsibilities should be separate. The latter should be a method on its own that can also be reused in https://gitlab.com/gitlab-org/gitlab/-/blob/26c2e6302994dfa99b8cb567704b94d29b416c9c/lib/gitlab/ci/build/context/build.rb#L42 as SSoT:

  module Ci::Deployable
    def expand_environment_name(environment)
      ExpandVariables.expand(environment, -> { simple_variables.sort_and_expand_all })
    end
  
    # At this stage the "expanded" prefix is irrelevant from a domain perspective
    def environment_name 
      return unless has_environment_keyword?
  
      if use_job_environment_for_expanded_environment_name? && job_environment&.expanded_environment_name.present?
        return job_environment.expanded_environment_name
      end
      
      metadata&.expanded_environment_name
    end
  end

  Then use explicitly expand_environment_name when we don't have job_environment persisted yet and use environment_name after a job is persisted.