Deploy N general agents with learnable prompt

[kongzii]

Now that we have

1. Agent that can learn with randomness (sometimes the agent even completely fails and just loops through the learning functions, other times it's super fast to get into trading)
2. It can save/load its prompt from DB
3. Viewer app

It would be interesting to have multiple instances of just_born agents and then watch and show to people how differently are they evolving their prompts and behaviour.

We'd need to:

1. Split MICROCHAIN_AGENT_OMEN in AgentIdentifier into multiple (e.g. MICROCHAIN_AGENT_OMEN_0, MICROCHAIN_AGENT_OMEN_1, etc)
2. Modify DeployableMicrochainAgent to allow select between them
3. Add their deployment
4. Modify the viewer app to switch between them

[evangriffiths]

why high?

• good for blog post (observe different behaviour, evidence of learning)
• good for debugging