add method to export small data set as numpy array, astropy table and/or panda dataFrame

[bombrun]

There are method to export to hdf5 or fits file but no method to export directly a selection (small enough) to a classical array or table format such as numpy array, astropy table and/or panda dataFrame.

ds.evaluate() might be part of the solution but it does not accept selection option.

[maartenbreddels]

@quiquinSP:

What I had in mind, are a few methods in the Dataset class like

• as_dataframe (for pandas)
• as_astropy_table
• as_numpy (a dict with numpy arrays)

It makes sense for them to take a 'selection' argument, see for instance Dataset.scatter, and the arguments column_names and virtual like export_hdf5, possibly a strings argument like Dataset.get_column_names(..). It may be convenient to give the as_numpy method an extra with_units=False argument to put in the units (see http://docs.astropy.org/en/stable/units/ ). Astropy tables can also take units, for pandas dataframe I am not sure.
You might want to check out in Dataset.scatter how I protect against converting too much data, or see https://github.com/maartenbreddels/vaex/blob/2754156da08fd4fad2555fdf0d85373ebae10a35/vaex/export.py#L291 how I protect against using too much memory, and similary in the gui https://github.com/maartenbreddels/vaex/blob/master/vaex/ui/main.py#L1070 (Just noticed that the gui code should be refactored to use the vaex.utils)
Please also include unittests, looking forward to see you PR!

[maartenbreddels]

Implemented in 7ca315b, with a short description here:
http://vaex.astro.rug.nl/latest/getting_data_in_vaex.html#getting-your-data-out
should be available in the next release