Consolidate Kimi providers under kimi while preserving moonshot compatibility

[AiCaptain357]

The dashboard calls this provider Kimi, but its models use moonshot/..., while kimi remains a separate hidden legacy entry. Both entries share the Moonshot executor and static model catalog. This leaves users with two names for the same API integration and maintainers with duplicate registrations to keep aligned.

Proposed direction

Make kimi the single canonical provider ID and advertised model prefix for this integration:

• Consolidate the current moonshot and legacy kimi registrations under kimi.
• Keep moonshot/... working as a backward-compatible alias.
• Migrate saved provider state while preserving connection IDs, credentials, per-connection endpoints, combos, and API-key access restrictions.
• Reconcile behavior differences explicitly and test existing configurations using either prefix.

Kimi Code and Kimi Web would remain separate integrations. This targets only the duplicate API-provider entries, including connections with custom endpoint overrides.

Why revisit this?

#7531 retained the hidden legacy entry for compatibility, and #8039 changed the moonshot entry’s display name to Kimi while deliberately keeping its ID, alias, and routing unchanged. I propose completing the consolidation while retaining compatibility through the old prefix. Freepik → Magnific (#10594) provides a precedent for a canonical-ID migration with a working legacy alias.

One kimi identity would make the provider name, model selection, and configuration consistent and remove the duplicate registration.

Would you support moving in this direction? I would like to prepare a focused migration and compatibility PR if this direction fits the project.