Documentation: add a process for cleaning up stale container images

[jayaddison]

Is your feature request related to a problem? Please describe.
During a recent ad-hoc cleanup of stale container images, some current (in-production) container images were mistakenly deleted, and this eventually caused the website outage described in #44.

We should determine and document a process to safely clean up stale container images.

Describe the solution you'd like
Documentation, including command-line examples, about how to remove stale container images as an operational process in the README.md file of this repository.

Describe alternatives you've considered
N/A

Additional context
Arises from prevention measures discussed in #45.

[jayaddison]

I'd suggest that we retain at least two images for each Kubernetes deployment:

• The image built from the same commit that the currently-running deployment was built from.
• The most recent image build that is not the currently-running deployment.

One complexity for the latter is that we recently began omitting timestamps from the container images themselves (e.g. openculinary/api@db144f4). We could however determine the next-most-recent image by referring to the source control history of the main branch (starting from the currently-deployed commit ID).