Message Table: Make it possible to search by Relevance

## What?
Add an option to sort messages in the **Message Table** by **Relevance (_score)** in addition to the existing timestamp-based sorting.

When users execute a search using Lucene query syntax — especially when applying boosting via the caret (`^`) operator (e.g., `error^5 OR warning`) — the result ranking is calculated internally by OpenSearch. However, the current message table primarily supports timestamp-based sorting, which prevents users from leveraging scoring-based ranking directly in the UI.

The proposed feature would:

- Add **“Relevance”** as a selectable sort option in the message table.
- Use the underlying `_score` field from OpenSearch.
- Clearly indicate in the UI when relevance sorting is active.

This would allow users to meaningfully use:
- Term boosting (`term^5`)
- Field boosting (`title:error^3`)
- Phrase boosting (`"database error"^10`)
- Multi-clause weighted queries

## Why?
Currently, users can technically use the caret (`^`) operator in Lucene queries to boost terms, but there is no visible effect in the UI unless results are sorted by relevance. Since most workflows default to timestamp sorting, boosting appears non-functional or misleading.

This impacts:

- Advanced search tuning
- Investigative workflows where prioritization matters
- User understanding of Lucene scoring capabilities

By enabling relevance-based sorting in the message table:

- Users can prioritize important signals over noise.
- Boosting becomes a practical and visible feature.
- Search behavior aligns more closely with OpenSearch capabilities.
- Power users gain better control over investigative workflows.