Add semantic caption output for accessibility

Problem/Motivation

The current template does not follow the correct semantic structure for image captions, as described in accessibility guidelines:

• https://stevenmouret.github.io/web-accessibility-guidelines/accessibilit...
• https://developer.mozilla.org/en-US/docs/Web/HTML/Reference/Elements/fig...

Captions should be wrapped in a <figure> element with a corresponding <figcaption> element to improve accessibility and provide meaningful HTML structure.

Steps to reproduce

1. Enable the module.
2. Enable the image caption option on an image field.
3. Inspect the rendered Twig output.
4. Caption is not wrapped in a semantic <figure>/<figcaption> structure.

Proposed resolution

1. Split caption output into legacy and semantic modes.
2. Enable legacy mode by default for existing users to preserve backward compatibility.
3. Provide a dedicated configuration form to switch between these modes for existing users.
4. Display warnings to users who are still using legacy mode that it is deprecated.
5. For new users, enable semantic mode by default.

Remaining tasks

N/A

User interface changes

• A new standalone form is available to switch between legacy and semantic modes.
• Warning messages are displayed in the formatter when legacy mode is enabled.
• Status report includes a warning if semantic mode is not active.

API changes

N/A

Data model changes

N/A