Feature: Support multiple custom URL/Key configurations with auto-switching

[AI-J-IN]

Problem

Problem

When selecting an AI model in Raven, the custom URL and API Key (the Custom type) only allows a single instance to be configured. It is not possible to configure multiple instances and enable auto-switching or load balancing between them.

Current Behavior

• In the model selection interface, the Custom type can only be configured with a single URL + Key pair
• It is not possible to add multiple custom endpoints, nor to auto-switch between them
• Switching endpoints requires manual reconfiguration

Impact

When users need to integrate with multiple custom API endpoints (e.g., multiple privately deployed model services, or backup Keys for failover), only one can be selected at a time. This is very inconvenient for multi-Key concurrency or auto-degradation scenarios.

Proposal

Proposal

• Support configuring multiple Custom URL + Key pairs
• Support auto-switching between these configurations (e.g., round-robin, or automatic fallback on failure)
• The model selection interface should display labels for multiple Custom configurations for easy identification

Alternatives considered

No response

Area

CLI / TUI

Additional context

No response