No space left on device when using images that are too big

[d2lam]

Currently, our VMs are using the default memory limit, which is 10GB. There are builds that use bigger images, and while pulling the docker image, it fails with error No space left on device

The pod just dies in the middle, and the build appears stuck. The status of the build remains as QUEUED.

We need to:

• Increase the memory limit/allow configuration
• Somehow reports back to the API that the build failed

[d2lam]

This was split into the following issues:
#757
#759
#758