Use file-like objects as parameters to train_* methods

[joshi-sh]

Currently, my workaround is to use tempfile as:

import tempfile
import fastText as ft
with tempfile.NamedTemporaryFile() as f:
    f.write(lines) 
    ft.train_supervised(input=f.name)

If I could instead pass a file-like object, it would be the single line ft.train_supervised(input=io.StringIO(initial_value=lines))

[Celebio]

Hi @joshi-sh ,
Thank you for your feedback.
We might think of an implementation based on this suggestion.

However we can't commit on the release date.
Thank you again.

Onur