[gitlab-org/gitlab] Update advanced search migration framework to conditionally use DOC_TYPE

### ROOT CAUSE The issue arises because the advanced search migration framework requires `DOC_TYPE` for all indexes, which is unnecessary for dedicated indexes (e.g., `vulnerability_reads`). This inefficiency stems from two main reasons: (1) the shared index (`main`) requires `DOC_TYPE` to distinguish multiple data types, and (2) `DOC_TYPE` is used to exclude validation records. However, dedicated indexes waste storage by including unused fields and prevent the creation of new indexes without `DOC_TYPE`. ### CODE FIX To conditionally use `DOC_TYPE` in the migration framework, modify the framework to check if the index is dedicated (i.e., handles a single document type). For dedicated indexes, omit `DOC_TYPE` during backfill/migration. For the shared index, retain `DOC_TYPE` for validation and type distinction. **Changes:** 1. **Index Configuration**: Add a flag (e.g., `requires_doc_type`) to index definitions to mark shared vs. dedicated indexes. 2. **Migration Framework**: Update the framework to use `DOC_TYPE` only when the index requires it. For dedicated indexes, skip validation records and `DOC_TYPE` entirely. **Example Code Snippet (Pseudocode):** ```ruby # In index definition class VulnerabilityReadsIndex < SearchIndex requires_doc_type false # Mark as dedicated end # In migration framework def migrate(index) if index.requires_doc_type # Use DOC_TYPE logic for shared indexes else # Skip DOC_TYPE for dedicated indexes end end ``` This change ensures dedicated indexes are created without `DOC_TYPE`, optimizing storage and enabling seamless migration.