After updating to Pandas 1.2.0, the below test in test_pandas is now failing

df = pytest.college_df
cdf = df.convert_dtypes()
cdf._repr_html_()
assert list(cdf.recommendation.keys()) == ["Correlation", "Distribution", "Occurrence"]

It raises TypeError: data type not understood when the cdf is displayed with _repr_html_.

This test works in Pandas 1.1 but fails in Pandas 1.2 .

In the documentation for convert_dtypes, we see:

Changed in version 1.2: Starting with pandas 1.2, this method also converts float columns to the nullable floating extension type.

This new change could be potentially causing the bug.