Add Unit-tests for prompts with expected outcomes

[Jgmedina95]

I talked about this with @SamCox822 a while ago. I'll add it here so I don't forget about it in the future.

As we move on and update and improve the system, there are minimal and medium prompts that we would expect to keep working precisely the same. These are the ones I expect to have in the unit test system.
Examples could be as minimal as:

• Download PDBID
• Make a plan for ONE-TWO-STEP-TASK
  or a bit more directed like:
• Use TOOL with PARAMETERS

So that when we change the system prompt, either adding a new tool, changing the output, or descriptions of existing tools, these minimal examples should remain working.

This will not be done with pytest, but maybe as 1) an internal tool that we could call ourselves to keep on check, 2) a recurrent system check that can also alarm us if something funny occurs with the APIs