WIP: Persist index data

[Zeegaan]

This is a working progress, but it aims to resolve #10

Problem

Index rebuilds can be slow because field values need to be calculated for every piece of content. This involves property value handlers, data collection services, and various transformations - all of which are computationally expensive when performed across thousands of documents.

Solution

Persist the calculated IndexField[] data to a database table (umbSearchDocuments) when content is indexed. During index rebuilds, this cached data can be reused instead of recalculating fields from scratch.

A useDatabase parameter controls this behavior:

• useDatabase: true (default) - Use cached fields from database if available
• useDatabase: false - Ignore cached data and recalculate fields fresh

When content types are modified (properties added/removed), the system automatically passes useDatabase: false to ensure stale cached data isn't used.

What's Been Done

Database Layer

• Added Document model for persisted index data (Models/Persistence/Document.cs)
• Added DocumentDto for database mapping
• Implemented IDocumentRepository and DocumentRepository with CRUD operations
• Added IDocumentService and DocumentService as the service layer
• Created database migration CustomPackageMigration to create the umbSearchDocuments table

Content Change Strategies

• Refactored DraftContentChangeStrategy to persist fields on index and use cached data on rebuild
• Refactored PublishedContentChangeStrategy with the same persistence pattern
• Added GetManyAsync for batch fetching documents (optimizes descendant processing)
• Updated IContentChangeStrategy.RebuildAsync signature to include useDatabase parameter

Index Rebuild Flow

• Updated IContentIndexingService.Rebuild to accept useDatabase parameter
• Modified RebuildIndexesNotificationHandler to pass useDatabase: false on content type changes and language deletion