added mistral-medium-latest

https://docs.mistral.ai/getting-started/models/models_overview/#api-versioning

Medium offers a reasonable balance between Small and Large in output and speed. Especially now when Large is sometimes is frequently over capacity.

This is my go to model for Mistral.

VoiceInk/Services/AIService.swift

@@ -114,6 +114,7 @@ enum AIProvider: String, CaseIterable {
         case .mistral:
             return [
                 "mistral-large-latest",
+                "mistral-medium-latest",
                 "mistral-small-latest",
                 "mistral-saba-latest"
             ]