Skip Pubchem live API test in CI (#1288)

Even with many retries and long delays, the test fails for almost every
PR. I don't know why this has been happening lately; perhaps Pubchem's
servers have become overloaded lately (possibly due to an increase in
the use of AI agents?). In any case, there seems to be no good solution
to stop flaky failures in CI except to simply skip the test in CI.

The approach implemented here tests an environment variable set by
GitHub in their runner environments. If that variable is set, the live
API test is skipped; otherwise, it's run as usual. This means that
developers in their normal local environments will run the tests as
usual, and don't have to do anything special. In addition, our CI
workflows don't need to change either.

---------

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Pavol Juhas <pavol.juhas@gmail.com>

Author: 3 people

src/openfermion/chem/pubchem_test.py

@@ -12,9 +12,11 @@
 
 """Tests for pubchem.py."""
 
+import os
+
 import numpy
-import pytest
 import pubchempy
+import pytest
 
 from openfermion.chem.pubchem import geometry_from_pubchem
 from openfermion.testing.testing_utils import module_importable
@@ -149,6 +151,11 @@ def test_water_2d(self, monkeypatch):
         with pytest.raises(ValueError, match='Incorrect value for the argument structure'):
             _ = geometry_from_pubchem('water', structure='foo')
 
+    # Skip if running in a CI environment.
+    @pytest.mark.skipif(
+        'CI' in os.environ,
+        reason='Skipping Pubchem API tests in CI to avoid failures due to busy servers.',
+    )
     @pytest.mark.flaky(retries=8, delay=30, only_on=[pubchempy.ServerBusyError])
     def test_geometry_from_pubchem_live_api(self):
         water_geometry = geometry_from_pubchem('water')