feat(internal): Automated test of demo

[planetf1]

POC of running a basic demo

• Runs the crewAI activity planner workflow (basic, does not use bee)
• Workflow launches Ollama only (not bee stack)
• Uses llama3.2:1b model
• no checking of output

Based on results of this (after discussion) could consider

• Generalizing to search for demos matching certain criteria
• Adding bee-stack (this might require a custom action, use of a composite compose, or k8s, perhaps KinD)
• how to define expected results

Other tests are also possible that stub out external calls to ollama

[maximilien]

There is still a failed check. Are you planning to solve or do you want me to?

[maximilien]

Don't add bee stack. Get the failed checks fixed. Thx

[planetf1]

Don't add bee stack. Get the failed checks fixed. Thx

I should clarify, the 'to consider' list is not for now (just input for discussion) This PR is just to understand the feasibility of automating one of the workflows that uses ollama that's been a bit problematic.

Will look at schema in separate PR.

[maximilien]

I am saying don't worry about "to consider" get this PR passing checks. And be done for now. Thx

[maximilien]

Parts are failing but will fix this in separate PRs. Let's merge this for now. Thx