[read_array_from_csv] add more flexibility when reading results

[fish-quant]

[read_array_from_csv] add more flexibility when reading results

[Henley13]

I see two different use cases :

• using a csv to store tabular data without header, but reading them as numpy ndarray matrix.
• the same, with a header, reading them as a dataframe.

If you just read and write raw tabular data without header, I could use numpy. If you want to keep headers, I should use pandas.

[Henley13]

Actually, I use the same function to write csv files (numpy arrays are saved with np.savetxt, without header, pandas objects are saved with data.to_csv, with header if it's requested by the user). To read the csv file, I define two different functions:

• bigfish.stack.read_dataframe_from_csv() to return a pandas object (with a header)
• bigfish.stask.read_array_from_csv() to return a numpy array (without header)

[Henley13]

Answer this issue in e0f15a4.