[enhancement]: Metadata/Prompt improvement

### Is there an existing issue for this?

- [x] I have searched the existing issues

### Contact Details

Xiaden#6585

### What should this feature add?

Enable user-defined checkpoint-specific suggested prompts to be loaded from checkpoint metadata and surfaced at the top of the Prompt Templates dropdown when a model is selected.

This would:

Keep checkpoint-relevant prompts close to their context

Reduce Prompt Template bloat from model-specific clutter

Enable quicker creative exploration using model-tuned phrasing

### Alternatives

I store model-specific prompt templates in a separate Obsidian vault

I could keep these in the general Prompt Templates system, but that quickly becomes overwhelming with too many checkpoint-specific entries that don’t generalize

This feature removes the need for external tools or cluttered templates by letting the app natively support scoped suggestions.

### Additional Content

Backend:

Add a new field to CheckpointConfigBase:

prompt_suggestions: Optional[List[str]] = Field(default_factory=list)

Stored in the database in the config JSON blob

Editable in the frontend (see below)

Frontend – Model Manager (Edit Checkpoint):

Section labeled "Suggested Prompts"

Four editable input fields:

Positive Prompt

Positive Style

Negative Prompt

Negative Style

Each with:

a "+" to add a new entry

an X-button in a dropdown to remove existing ones

Frontend – Text-to-Image & Image-to-Image:

When a checkpoint is selected:

These suggestions appear at the top of the Prompt Templates dropdown

Clearly labeled as “Checkpoint Suggestions”

Other user-created/global templates remain below

Optional enhancements:

[ ] Integrate into workflows

[ ] Pull suggestions from external sources (e.g., CivitAI or internal prompt database.

(Again, GPT cleaning up my sloppy writing...)