Fixes #22

Currently I have the following added to all of my workflow files using the deployerphp action for our self hosted runners:

- name: Clean Up SSH Agent
  if: always()
  run: |
    killall ssh-agent
    rm ~/.ssh/config

This works fine, however recently we had started having multiple runner instances on each of our runner servers and conflicts began to arise with my solution due to the usage of killall

While I could have improved killing off the ssh-agent, I've decided to take a crack at implement a way of handling the ssh-agent process a little more nicely within the action iteself and have ended up with this.

There is probably a slightly better way of doing this via a proper setting, however I was aiming for backwards compatibility where possible.

What this PR does:

1. Spawns the SSH agent without specify a socket path, ensuring a unique socket path for each runner.
2. Adds a new option to toggle if the action should disable strict host checking if no known hosts are provided.
3. Adds a cleanup task that ensures all SSH keys are removed from the SSH Agent.
4. Ensures the SSH is shutdown on the runner server.

Why make these changes:

1. Improves supports for self hosted runners while also adding a bit of extra security for those using GitHub hosted runners.
2. Allows for the action to still handle the SSH related tasks for the end user without actually touching anything SSH related if necessary.
3. Probably doing things "the right way", so to speak, in that we're cleaning up after ourselves as part of the action.