equality comparison with a scalar is slow for category (performance regression) #23814
Description
Are the following 2 ways to compare a series to a scalar equivalent (ignore missing values)? I have to write the hard way in order to take advantage of the category properties.
```python
x = pd.Series(list('abcd') * 1000000).astype('category')
%timeit x == 'a'
# 10 loops, best of 3: 25.2 ms per loop
%timeit x.cat.codes == x.cat.categories.get_loc('a')
# 1000 loops, best of 3: 750 µs per loop
```