Refactor Transformer for batch `ingest->process` workflow

• Leverage schema service for authoritative source on property types
  • Combine with inferred types (assumed string) for maximum coverage and compatibility
• Introduce loadInBatch capability which enables a substantially more performant and memory-conscious workflow for ingestion and processing (stages 1 & 2 of Transformer)
  • batchSize is configurable for fine-tuning the performance to an operator's environment
• Docs updated to reflect new capabilities
• Minor bug fixes to Provider