This PR will do the following:

• move tasks for rolling start and stop into separate files
• include the new files in the rolling upgrade tasks file
• replace the simple "Restart Elasticsearch handler" with a new one that includes a file which itself includes the new task files

The idea is to make every single restart of Elasticsearch a rolling one.

Status:

• Review the split up task lists and search for potential problems. Especially with when clauses. (For mow it's just moved code
• Re-implement the systemctl daemom reload that went missing when replacing the restart task
• Run real life tests

fixes #343

The idea to use include files that can be reused comes from @ivareri