feat: add 2026_may folder and nppes data example

[asenetcky]

---

Title: Add May 2026 Example: Handling Large NPPES Data with DuckDB & Parquet

Summary
This PR adds a new module for the May 2026 R User Group session focusing on efficient ingestion and processing of large datasets. The example uses the National Plan and
Provider Enumeration System (NPPES) "Full Replacement File" (~11GB CSV) to demonstrate how to bypass memory limitations using DuckDB and Parquet.

Key Changes

• New Folder: 2026_may/working-with-large-nppes-data/
• Ingestion Script (ingest-nppes.R):
  • Automates scraping the CMS website for the latest monthly NPPES download link.
  • Downloads and extracts the data using httr2 and utils::unzip.
  • Leverages DuckDB to "stream" the massive CSV from disk, filtering for Connecticut (CT) providers without loading the full file into RAM.
  • Exports the filtered results to a compressed Parquet file for high-performance downstream analysis.
• Documentation (README.md): Detailed walkthrough explaining the "Out of Memory" problem and why the DuckDB + Parquet architecture is a superior solution for
  large-scale CSV work.
• Environment Management (renv.lock): Full lockfile to ensure reproducibility across different machines.

Why this is valuable
Health care data (like NPPES) is often too large for standard read_csv() calls on consumer-grade hardware. This contribution provides the group with a template for
"out-of-memory" data engineering in R, making these datasets accessible to more users.

How to Test

1. Clone the branch and navigate to 2026_may/working-with-large-nppes-data/.
2. Run renv::restore() to install dependencies.
3. Source ingest-nppes.R.
4. Verify that data/nppes/nppes_ct/ is created containing the filtered Parquet dataset.

---