Remove yum-cron updates from PMM Server

Problem

yum-cron that runs every 10 minutes is conflicting with the upgrade process: in rare cases when both yum-cron and pmm-update (ansible) query the BerkelyDB-based local yum registry, the latter becomes broken and irrecoverable via standard OS tools.

Solution

Remove yum-cron from PMM-Server as it does not solve the problem of applying the security updates.

Note: the newest version of rpm  tool has already migrated from BerkleyDB to sqlite, which makes the tool more reliable. CentOS 7 still does not have it available.

Benefits

1. More reliable upgrade process
2. Less risks to run processes which we do not control and which could potentially break the system.

How to test - AMI

1. Run a PMM Server instance based on an the AMI image ami-0941ae4fb3b8be137 with two volumes: 8GB and 100GB
2. Manually resize the /srv (/dev/sdb) disk from AWS interface, wait for 5 minutes, ssh to the instance and verify that the disk is resized (see the attached screenshot)
3. run `supervisorctl status` and verify that cron is not in the list of processes managed by supervisord
4. run `sudo systemctl status crond` to verify that the crond service is running

NOTE: AMI can only be tested when we merge the PR in pmm-server repo.

How to test - Docker

1. Run a PMM Server instance based on image perconalab/pmm-server-fb:PR-2459-65f7a1b
2. ssh to the instance, verify by running `ps aux | grep cron` that cron is not running
3. run `supervisorctl status` and verify that cron is not in the list of processes managed by supervisord