fix: use Duration strings for Polars object_store S3 timeout options

Polars' native Rust reader passes storage_options through to the object_store
crate, which requires Duration strings (e.g. '10s', '120s') rather than plain
numeric second values ('10', '120').

pyiceberg's PyArrowFileIO uses float(value) for the same keys, so the two callers
need different formats. Split into separate dicts: pyiceberg_s3_properties (plain
numbers for GlueCatalog) and polars_s3_storage_options (duration strings for
pl.scan_iceberg).

Fixes: object-store error: Generic Config error: failed to parse "120" as Duration

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: blarghmatey

packages/ol-orchestrate-lib/src/ol_orchestrate/lib/glue_helper.py

@@ -106,22 +106,26 @@ def get_dbt_model_as_dataframe(database_name: str, table_name: str) -> pl.LazyFr
         KeyError: If the table metadata doesn't contain the expected fields
         boto3 exceptions: If the AWS Glue API call fails
     """
-    # s3.connect-timeout and s3.request-timeout apply to pyiceberg's PyArrowFileIO
-    # (used for manifest reads). They are also mapped by Polars 1.40+ to
-    # object_store's `connect_timeout` and `timeout` for its native Rust S3 reader,
-    # but only when passed explicitly as storage_options to pl.scan_iceberg().
-    # Without them, Polars' Tokio runtime accumulates CLOSE_WAIT S3 connections
-    # that block process exit indefinitely.
-    s3_storage_options = {
+    # pyiceberg's PyArrowFileIO expects timeout values as plain numeric seconds strings.
+    pyiceberg_s3_properties = {
         "s3.region": "us-east-1",
         "s3.connect-timeout": "10",
         "s3.request-timeout": "120",
     }
+    # Polars 1.40+ maps s3.connect-timeout / s3.request-timeout to object_store's
+    # `connect_timeout` / `timeout` for its native Rust S3 reader, preventing
+    # CLOSE_WAIT connections from blocking Tokio runtime shutdown at process exit.
+    # object_store requires Duration strings (e.g. "10s"), not bare integers.
+    polars_s3_storage_options = {
+        "s3.region": "us-east-1",
+        "s3.connect-timeout": "10s",
+        "s3.request-timeout": "120s",
+    }
     glue = GlueCatalog(
         "default",
         client=boto3.client("glue", region_name="us-east-1"),
-        **s3_storage_options,
+        **pyiceberg_s3_properties,
     )
     table = glue.load_table(f"{database_name}.{table_name}")
 
-    return pl.scan_iceberg(table, storage_options=s3_storage_options)
+    return pl.scan_iceberg(table, storage_options=polars_s3_storage_options)