Add an APPEND option to buildParquetCopySQL

buildParquetCopySQL() gains an `append` argument that emits the DuckDB `APPEND`
copy option. A `COPY ... PARTITION_BY` into a non-empty target otherwise fails
with "Directory ... is not empty"; `APPEND` writes new files into the existing
hive-partitioned directory (with unique names, preserving the files earlier
parts wrote), which is what a coord-array append needs. Default is FALSE, so the
flat/lazy and fresh-write paths are unchanged; DuckDBArray's partitioned writer
passes append = TRUE for the append case.

Bump version to 0.99.6; update NEWS.

Author: aboyounp-gne

DESCRIPTION

@@ -1,5 +1,5 @@
 Package: DuckDBDataFrame
-Version: 0.99.5
+Version: 0.99.6
 Date: 2026-07-22
 Title: DuckDB-Backed DataFrame and Table Structures
 Description:

NEWS.md

@@ -1,3 +1,12 @@
+# DuckDBDataFrame 0.99.6
+
+## New features
+
+- `buildParquetCopySQL()` gains an `append` argument that emits the DuckDB
+  `APPEND` copy option. This lets a `PARTITION_BY` write add new files to an
+  existing (hive-partitioned) directory instead of failing "directory is not
+  empty", which is what a coord-array append needs (DuckDBArray).
+
 # DuckDBDataFrame 0.99.5
 
 ## Bug fixes

R/parquet-io.R

@@ -348,7 +348,7 @@ quoteSQLColumns <- function(conn, cols) {
 #' @rdname parquet-io
 buildParquetCopySQL <-
 function(query_sql, target_path, order_cols = NULL, partition_by = NULL,
-         row_group_size = NULL)
+         row_group_size = NULL, append = FALSE)
 {
     order_clause <- if (!is.null(order_cols) && length(order_cols) > 0L) {
         sprintf(" ORDER BY %s", paste(order_cols, collapse = ", "))
@@ -369,6 +369,9 @@ function(query_sql, target_path, order_cols = NULL, partition_by = NULL,
         options <- c(options,
                      sprintf("PARTITION_BY (%s)", paste(partition_by, collapse = ", ")))
     }
+    if (isTRUE(append)) {
+        options <- c(options, "APPEND")
+    }
 
     sprintf(
         "COPY (%s%s) TO '%s' (%s)",