Initialize Speedml with pandas dataframe rather than .csv

Not all data comes in .csv or .json files.  It would be fantastic and helpful to allow for initializing the process with a dataframe, matrix or series of data rather than requiring files.  For instance, 

```python
import pandas as pd
from speedml import Speedml

df = pd.read_excel('../somefile')
sml = Speedml(data = df, target='column name')

```