Changed the 'enable_rsyslog_config' variable from false to true
This should be the default behavior for all nodes unless explicitly turned off by the developer.

Adds fail2ban filter and jail configs

Created symlinks for `slurm.conf`, `gres.conf`, `slurmdbd.conf`, and `cgroup.conf` in `/etc/slurm`.
To ensure consistency with the current production environment where these symlinks already exist.

We moved cheaha.yml playbook to run during deploy time

previously it reflected a placeholder but it actually is the value in file from build that we're looking for and replacing in deploy phase

We lookup the value of the placeholder (RewriteCond %{HTTP:REMOTE_USER}  \’^(.+)$\’) in this case in the apache conf during build and replace it with the rewrite conditions during deploy phase.

The home dir for slurm user is available as a mount from master so there
is no need to create a local home dir for it.

Follow the same approach in ssl_cert role, to avoid missing host key issue during deploy

pam_slurm config denies ssh if there is no slurm job on the compute node
A login node is not expected to run jobs so we removed this config from
the login playbook

Change the references to compute.yml in the packer template to use
login.yml

So that we can have same key names for both dicts used in autofs.yml and fstab.yml roles

The values used in prod will be defined at runtime so replacing them
with default values

Closes rc/hpc-factory#173

Closes rc/hpc-factory#175