Rework AI model selection bulk update UI

What does this MR do and why?

Reworks the bulk update UI in the AI model selection admin settings, for both the namespace (model_selection) and instance (instance_model_selection) settings.

The old per-row Apply to all button obscured what it actually did. It was not clear which model would be applied, or why it would only update some records. This MR makes that behavior explicit: a single section-level Bulk update dropdown lets you pick the model to apply, and each model is applied only to the sub-features it is compatible with instead of requiring it to be valid for all of them.

• Replaces the per-row "Apply to all" button with a single section-level Bulk update dropdown plus a Reset to default action.
• The dropdown offers the union of models selectable across a section's sub-features and applies each model only to the sub-features it is compatible with, rather than requiring a model to be valid for every sub-feature.
• Replaces the collapsible section wrapper with a flat section header that hosts the bulk update control.
• Adds a compact inline (non-block) mode to the shared model dropdown for this use.

References

N/A

Screenshots or screen recordings

Admin Area (Self-Managed)

Before	After

Group level (Saas)

Before	After

How to set up and validate locally

1. Enable model selection for a group or instance so the admin settings render.
2. Open the AI model selection settings page.
3. In a section with more than one sub-feature, use the "Bulk update" dropdown to apply a model, and confirm it applies only to compatible sub-features (toast reports the count updated).
4. Use "Reset to default" and confirm all sub-features return to their default model.

MR acceptance checklist

Evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.