feat: add per-modality default model configuration

[raylanlin]

Summary

Allow users to set default models in config.json for each modality, avoiding the need to pass --model on every command.

New Config Keys

Key	Hyphen alias	Default
default_text_model	default-text-model	MiniMax-M2.7
default_speech_model	default-speech-model	speech-2.8-hd
default_video_model	default-video-model	MiniMax-Hailuo-2.3
default_music_model	default-music-model	key-type-based

Resolution Priority

1. Explicit --model flag (highest)
2. Config file default
3. Hardcoded fallback (lowest)

Special Cases

• Video: Auto-switch flags (--lastFrame, --subjectImage) still take priority over config defaults.
• Music Cover: defaultMusicModel only applies if the value is a valid cover model name (music-cover / music-cover-free), otherwise the key-type-based default is preserved.

Usage

# Set defaults
mmx config set --key default-text-model --value MiniMax-M2.7-highspeed
mmx config set --key default-music-model --value music-2.6

# Use without --model
mmx text chat --message "Hello"
mmx music generate --prompt "Upbeat piano" --instrumental

# --model still overrides
mmx text chat --model MiniMax-M2.7 --message "Hello"

Tests

• 97 tests pass (24 new tests added)
• Coverage: config set/show, text chat, speech synthesize, video generate, music generate, model resolution logic, music cover isolation