cleanup redundant logic to initialize manta credentials

[Prakash Surya]

In a few different scripts there exists logic like this:

log_must mkdir -p $HOME/.ssh
log_must chmod 700 $HOME/.ssh

log_must vault_read_manta_private_key > $HOME/.ssh/id_rsa
log_must chmod 400 $HOME/.ssh/id_rsa

log_must vault_read_manta_public_key > $HOME/.ssh/id_rsa.pub
log_must chmod 644 $HOME/.ssh/id_rsa

log_must test -d "$OPENZFSCI_DIRECTORY"
log_must cd "$OPENZFSCI_DIRECTORY/ansible"

log_must cat > inventory.txt <<EOF
$HOST ansible_ssh_user="$USER" ansible_ssh_pass="$PASS"
EOF

log_must cat > playbook.yml <<EOF
---
- hosts: localhost
  tasks:
    - wait_for:
        host: $HOST
        port: 22
        state: started
        timeout: 1800
EOF

log_must ansible-playbook -vvvv -i inventory.txt playbook.yml >&2

This is essentially copy/pasted into the scripts that need to interact with manta. We should remove the code duplication and instead have a shell library file that would provide a common function to perform this logic; then all scripts that needed to perform these steps would just use the function.

The same goes for the part that uses ansible to wait for port 22; rather than copy/pasting, we should have a way for the scripts to share this logic.