[AppEx AI] Integrate pre-configured Kibana AI Connector to the O11y and Search GenAI powered functionality

[YulNaumenko]

Currently solutions AI Assistants are using three different GenAI connector types, which helps to integrate directly with OpenAI, Gemini and Bedrock.
The long term vision is to migrate from the explicit connectors implementation to the usage of the generic .inference connector type #189027, which is integrated with Elasticsearch Inference API. This connector supports multiple LLMs integrations (including OpenAI, Gemini, Bedrock and much more), but due to the big scope of the complete adoption and migration path to .inference connector, the scope was reduced to the MVP integration which includes the usage of the pre-configured connector with EIS service.

In 8.18 there is a plan to provide Elastic Default LLM experience, which will be exposed within Kibana .inference connector type.
Requirements:

• add .inference pre-configured connector instance to the list of the available connectors instances and select/use it by default (if the connector selection was not changed)
• prohibit the create/edit/delete of the new .inference connector instances.
• use sub action for _unified completion introduced by [AI Connector] Kibana inference connector should support openai compatible schema for completion task #202621

[pgayvallet]

Duplicate of #199082, probably?

Note that

add .inference pre-configured connector instance to the list of the available connectors instances and select/use it by default (if the connector selection was not changed)

Seems out of the scope for appex-ai-infra, that logic is living in the o11y assistant's code.

prohibit the create/edit/delete of the new .inference connector instances.

Not sure this is being relevant here, as it seems unrelated to supporting the inference connector?

[elasticmachine]

Pinging @elastic/unified-observability (Team:Observability)

[YulNaumenko]

Implemented by #199082