fix: signal handling for Concourse 7.5

[blgm]

There seems to be a change (unidentified as yet) in Concourse 7.5 that
results in no longer being able to stop or time-out once Terraform is
running. This change introduces a wrapper around the Terraform command
that catches signals and kills the Terraform process group if one is
detected.

Resolves #163

Notes:

• I haven't run the integration tests as they were tricky to set up, and I don't think they would catch errors in the changes that I made. I have however run this in my pipeline for a few days
• There's an argument that rather than using fmt.Printf() for logging, logger.Logger should be used instead. But I found that in order to pass logger.Logger then I would have to make a lot of changes in a lot of layers. I'm happy making this change if that's preferred.
• I did consider merging client.terraformCmd() into the new runner package to create a "tfrunner" package, but it looked like it would be a bigger change and wasn't clearly a better implementation.

[tinygrasshopper]

hey @ljfranklin, any thoughts on this PR.

We are currently using a forked version of the resource, it would be great to get back to using main.

[ljfranklin]

Thanks for the PR @blgm! The fix is now available in ljfranklin/terraform-resource:latest and ljfranklin/terraform-resource:1.0.10 images.