Align docs and behavior: address purpose, empty results, validation

[frankfarach]

Summary

• Normalize address_purpose inputs to the API’s uppercase values and add test coverage.
• Return a typed npi_results object for zero‑record searches.
• Vectorize npi_is_valid() to accept multiple NPIs.
• Clarify taxonomy fallback behavior in npi_summarize docs and README.
• Add Author/Maintainer fields to satisfy R CMD check.

Changes

• npi_search() accepts case‑insensitive address_purpose and normalizes to LOCATION/MAILING/PRIMARY/SECONDARY.
• npi_search() returns a structured empty npi_results instead of bare tibble().
• npi_is_valid() now accepts vectors and returns a logical vector.
• Docs updated to note taxonomy fallback when no primary is marked.
• NPI extraction coerced to integer to keep binding stable.

Testing

• R CMD build .
• R CMD check npi_0.2.0.9000.tar.gz

Notes

• No behavior changes beyond the described alignments; existing pipelines should remain compatible.