Add a set of basic functions for benchmarks and experiments

[ema-pe]

Currently, three functions (ocr, shasum, and figlet) are deployed on the FaaS platform. We should add more basic functions to allow more experiments and benchmarks, each based on different characteristics (for example, one should be CPU-bound or IO-bound). A subset of these functions should also cover ML-based applications, using some pre-defined models. These functions can be created from scratch, or better yet, we can find a curated set somewhere that we can use.

[ema-pe]

For the WatchEDGE project, we need to deploy a set of functions in the proof-of-concept that involve ML inference on images.

[ema-pe]

While creating a custom OpenFaaS function on the local registry, I got this error:

$ faas-cli deploy --image=localhost/image-classification --name=image-classification --label dfaas.maxrate=100

Unexpected status: 400, message: the Community Edition license agreement only allows public images


Function 'image-classification' failed to deploy with status code: 400

This is another reason to move away from OpenFaaS (see #42).

[ema-pe]

With commit 2f1c04f, an ML inference function has been added to the DFaaS repository. After this brief experiment, it’s better to pause for now until issue #42 is resolved, as it significantly slows down development and deployment.

[ema-pe]

I also tried using the Function CRD as described in this article, but it isn't available in OpenFaaS CE. Therefore, we are forced to use faas-cli instead.