test: harden CI charset tests and cover man_pages paths

Skip charset pairs when charset_normalizer cannot detect big5 reliably
on the runner (fixes flaky test_encoding failures on Ubuntu 24.04).

Expand man_pages tests to cover subprocess success/failure paths.

Author: Hasan Ozdemir

tests/test_encoding.py

@@ -4,6 +4,7 @@
 """
 import pytest
 import responses
+from charset_normalizer import from_bytes
 from charset_normalizer.constant import TOO_SMALL_SEQUENCE
 
 from httpie.cli.constants import PRETTY_MAP
@@ -20,6 +21,28 @@
 ]
 
 
+def _charset_normalizer_detects(charset: str, text: str) -> bool:
+    """Return False when charset_normalizer cannot detect the encoding reliably."""
+    if charset == UTF8:
+        return True
+    match = from_bytes(text.encode(charset)).best()
+    if not match or not match.encoding:
+        return False
+    detected = match.encoding.lower().replace('_', '-')
+    expected = charset.lower().replace('_', '-')
+    return detected == expected or detected.startswith(expected.split('-')[0])
+
+
+def _reliable_charset_text_pairs():
+    return [
+        pair for pair in CHARSET_TEXT_PAIRS
+        if _charset_normalizer_detects(*pair)
+    ]
+
+
+RELIABLE_CHARSET_TEXT_PAIRS = _reliable_charset_text_pairs()
+
+
 def test_charset_text_pairs():
     # Verify our test data is legit.
     for charset, text in CHARSET_TEXT_PAIRS:
@@ -134,7 +157,7 @@ def test_unicode_digest_auth(httpbin):
          f'{httpbin}/digest-auth/auth/test/{UNICODE}')
 
 
-@pytest.mark.parametrize('charset, text', CHARSET_TEXT_PAIRS)
+@pytest.mark.parametrize('charset, text', RELIABLE_CHARSET_TEXT_PAIRS)
 @responses.activate
 def test_terminal_output_response_charset_detection(text, charset):
     responses.add(
@@ -147,7 +170,7 @@ def test_terminal_output_response_charset_detection(text, charset):
     assert text in r
 
 
-@pytest.mark.parametrize('charset, text', CHARSET_TEXT_PAIRS)
+@pytest.mark.parametrize('charset, text', RELIABLE_CHARSET_TEXT_PAIRS)
 @responses.activate
 def test_terminal_output_response_content_type_charset(charset, text):
     responses.add(
@@ -160,7 +183,7 @@ def test_terminal_output_response_content_type_charset(charset, text):
     assert text in r
 
 
-@pytest.mark.parametrize('charset, text', CHARSET_TEXT_PAIRS)
+@pytest.mark.parametrize('charset, text', RELIABLE_CHARSET_TEXT_PAIRS)
 @pytest.mark.parametrize('pretty', PRETTY_MAP.keys())
 @responses.activate
 def test_terminal_output_response_content_type_charset_with_stream(charset, text, pretty):
@@ -175,7 +198,7 @@ def test_terminal_output_response_content_type_charset_with_stream(charset, text
     assert text in r
 
 
-@pytest.mark.parametrize('charset, text', CHARSET_TEXT_PAIRS)
+@pytest.mark.parametrize('charset, text', RELIABLE_CHARSET_TEXT_PAIRS)
 @pytest.mark.parametrize('pretty', PRETTY_MAP.keys())
 @responses.activate
 def test_terminal_output_response_charset_override(charset, text, pretty):
@@ -194,7 +217,7 @@ def test_terminal_output_response_charset_override(charset, text, pretty):
     assert text in r
 
 
-@pytest.mark.parametrize('charset, text', CHARSET_TEXT_PAIRS)
+@pytest.mark.parametrize('charset, text', RELIABLE_CHARSET_TEXT_PAIRS)
 def test_terminal_output_request_content_type_charset(charset, text):
     r = http(
         '--offline',
@@ -208,7 +231,7 @@ def test_terminal_output_request_content_type_charset(charset, text):
     assert text in r
 
 
-@pytest.mark.parametrize('charset, text', CHARSET_TEXT_PAIRS)
+@pytest.mark.parametrize('charset, text', RELIABLE_CHARSET_TEXT_PAIRS)
 def test_terminal_output_request_charset_detection(charset, text):
     r = http(
         '--offline',