Optionally disable verify_mapping_compatibility in pandas_to_eland append

[stevedodson]

When converting pandas DataFrames and appending to ES index, the current implementation is strict and enforces compatibility between the pandas DataFrame and the ES mapping.

A use case is multiple pandas DataFrames are appended to an ES index. For example,

df1 = pd.DataFrame(data={'key': ['test'], 'value1': [10] })
df2 = pd.DataFrame(data={'key': ['test'], 'value2': [20] })

ed.pandas_to_eland(df1, es_client, es_index, es_dropna=True, 
                               es_if_exists='append', use_pandas_index_for_es_ids=False)

# currently throws ValueError
ed.pandas_to_eland(df2, es_client, es_index, es_dropna=True, 
                               es_if_exists='append', use_pandas_index_for_es_ids=False)

It can be useful to disable this strict behaviour so new dynamic mappings can be created in the ES index. This would result in:

ed.DataFrame(es_client, es_index)

    key  value1  value2
0  test    10.0     NaN
1  test     NaN    20.0

Therefore, an option is to add an argument such as verify_mapping_compatibility: False