fix: suppress subprocess UTF-8 decode warnings in test runners

Add errors="replace" to subprocess.run() calls in test_regression.py,
test_snapshot.py, and test_e2e_pipeline.py. The validator emits
Windows-1252 bytes in its stderr color codes; replacing undecodable
bytes prevents PytestUnhandledThreadExceptionWarning noise without
affecting test correctness.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: amar-python

tests/snapshots/basic_expected_skip.csv

@@ -0,0 +1,3 @@
+id,name,_skip_reason
+empty row — all values blank
+2,"column mismatch — expected 2, got 1"

tests/snapshots/basic_expected_valid.csv

@@ -0,0 +1,3 @@
+id,name
+1,Alice
+3,Bob

tests/test_e2e_pipeline.py

@@ -60,6 +60,7 @@ def test_happy_path_two_valid_rows(self):
             r = subprocess.run(
                 [sys.executable, str(VALIDATOR)],
                 env=env, capture_output=True, text=True,
+                encoding="utf-8", errors="replace",
             )
             self.assertEqual(r.returncode, 0, r.stderr)
             valid_lines = valid_csv.read_text(encoding="utf-8").strip().splitlines()
@@ -87,6 +88,7 @@ def test_mixed_rows_splits_correctly(self):
             r = subprocess.run(
                 [sys.executable, str(VALIDATOR)],
                 env=env, capture_output=True, text=True,
+                encoding="utf-8", errors="replace",
             )
             self.assertEqual(r.returncode, 0, r.stderr)
             valid_lines = valid_csv.read_text(encoding="utf-8").strip().splitlines()
@@ -112,6 +114,7 @@ def test_all_invalid_rows_exit_nonzero(self):
             r = subprocess.run(
                 [sys.executable, str(VALIDATOR)],
                 env=env, capture_output=True, text=True,
+                encoding="utf-8", errors="replace",
             )
             self.assertEqual(r.returncode, 1)
             self.assertIn("No valid rows", r.stderr)
@@ -133,6 +136,7 @@ def test_output_directory_is_created_automatically(self):
             r = subprocess.run(
                 [sys.executable, str(VALIDATOR)],
                 env=env, capture_output=True, text=True,
+                encoding="utf-8", errors="replace",
             )
             # Either succeeds and created the dir, or fails with a clear message
             if r.returncode == 0:

tests/test_regression.py

@@ -25,6 +25,7 @@ def _run(env: dict) -> subprocess.CompletedProcess:
         capture_output=True,
         text=True,
         encoding="utf-8",
+        errors="replace",
     )
 
 

tests/test_snapshot.py

@@ -44,6 +44,7 @@ def _run_validator(csv_content: bytes, table_name: str = "people") -> tuple[str,
             capture_output=True,
             text=True,
             encoding="utf-8",
+            errors="replace",
         )
         valid_text = valid_csv.read_text(encoding="utf-8") if valid_csv.exists() else ""
         skip_text  = skip_csv.read_text(encoding="utf-8")  if skip_csv.exists()  else ""