Test loading JLSO files using Python's bson library

[rofinn]

This should ensure that:

1. We aren't using custom Julia specific types in the header
2. Demonstrate how to extract raw data from from a JLSO file outside of this library

Snippet:

using JLSO

julia> JLSO.save("python-test.jlso", :scores => Int64[1, 2, 3, 4]; format=:bson, compression=:none)

import bson
import struct

f = open("python-test.jlso", "rb")
fbytes = f.read()
header = bson.loads(fbytes)

obj_names = header['data'][1][0]

# Obviously you could generalize this to iterate over the `obj_names`
obj_nbytes = struct.unpack('l', header['data'][1][1]['data'])
obj_pos = struct.unpack('l', header['data'][1][2]['data'])

# Example extracting a bson data object
f.seek(obj_pos[0])
bson_data = bson.loads(f.read(obj_nbytes[0]))
scores = struct.unpack('llll', bson_data['data'][1][0]['data'])

f.close()