fix: Fixed bug in TabularDataset.column_names (#590)

Fixes #589

The `end` parameter of the `blob.download_as_bytes` function is inclusive, not exclusive.

> There are 2 hard problems in computer science: cache invalidation, naming things, and off-by-1 errors.

Co-authored-by: gcf-merge-on-green[bot] <60162190+gcf-merge-on-green[bot]@users.noreply.github.com>

google/cloud/aiplatform/datasets/tabular_dataset.py

@@ -150,7 +150,7 @@ def _retrieve_gcs_source_columns(
 
             while first_new_line_index == -1:
                 line += blob.download_as_bytes(
-                    start=start_index, end=start_index + increment
+                    start=start_index, end=start_index + increment - 1
                 ).decode("utf-8")
 
                 first_new_line_index = line.find("\n")