Convert nan to 0 in avg_num_tokens()

[hungcs]

avg_num_tokens() should always return an int, but it's possible for it to return a nan.

titanic_simple_df = pd.DataFrame(
    {
        "PassengerId": [1], "Survived": [0], "Pclass": [3], "Name": ["Braund, Mr. Owen Harris"],
        "Sex": ["male"], "Age": [22.0], "SibSp": [1], "Parch": [0], "Ticket": ["A/5 21171"], "Fare": ["7.25"],
        "Cabin": [None], "Embarked": ["S"], "split": [0]
    }
)

>>> cab = titanic_simple_df["Cabin"]
>>> Series(cab).str.split().str.len()
0    None
dtype: object
>>> Series(cab).str.split().str.len().mean()
nan

get_dataset_info_from_source(table_source).to_dict()\n  File 
"/home/ray/anaconda3/lib/python3.8/site-packages/ludwig/automl/base_config.py", 
line 189, in get_dataset_info_from_source\n    avg_words = 
source.get_avg_num_tokens(field)\n  File 
"/home/ray/src/predibase_engine/sources/dask_util.py", line 46, in 
get_avg_num_tokens\n    return avg_num_tokens(self.sample)\n  File 
"/home/ray/anaconda3/lib/python3.8/site-packages/ludwig/automl/utils.py", line 
61, in avg_num_tokens\n    avg_words = 
round(Series(unique_entries).str.split().str.len().mean())\nValueError: cannot 
convert float NaN to integer (type: ErrorResponse, retryable: true)')

[github-actions]

Unit Test Results

6 files  ±       0  6 suites  ±0   29s ⏱️ - 2h 7m 25s
1 tests  - 2 784  0 ✔️  - 2 750  0 💤  -   35  0 ❌ ±0  1 🔥 +1 
6 runs   - 8 349  0 ✔️  - 8 246  0 💤  - 109  0 ❌ ±0  6 🔥 +6 

For more details on these errors, see this check.

Results for commit bb3a870. ± Comparison against base commit 55b7672.

♻️ This comment has been updated with latest results.

[tgaddair]

Can you add a unit test?