Added provider-level batching for selected arXiv and NCBI operations
to reduce unnecessary live-service requests while preserving existing
public return shapes.
Added batched NCBI support for PMID, PMCID, and DOI metadata,
linked-identifier lookup, existence checks, and supported identifier
conversions.
Moved batch execution into the unary and binary dispatch engines so
batching is handled consistently through internal dispatchers rather
than exported API functions.
Added package-managed throttling for arXiv, NCBI, and Europe PMC
requests, with user-configurable rate-limit options.
Added a provider-etiquette vignette documenting batching,
throttling, live-service behavior, and relevant user options.