Parallelize index metadata migration in EsMetadataMigrationPipeline

[AndreKurait]

Description

EsMetadataMigrationPipeline.migrateAll() currently migrates index metadata sequentially using concatMap. Since index creation is independent across indices, this could be parallelized for better performance on clusters with many indices.

Current Behavior

Flux.fromIterable(indices)
    .concatMap(indexName -> migrateIndexMetadata(indexName).thenReturn(indexName))

Proposed Change

Replace concatMap with flatMap and a bounded concurrency:

Flux.fromIterable(indices)
    .flatMap(indexName -> migrateIndexMetadata(indexName).thenReturn(indexName), 5)

A default concurrency of 5 should be safe since index creation operations are independent.

Context

Identified during PR review of the RfsPipeline refactor. Kept sequential to match existing behavior and limit PR scope.

[sumobrian]

This is an example of a ticket that would go in Jira versus Github