Remote creation of numpy arrays is slow on large lists

[pilcru]

(From email)

The code below takes 5-10 seconds to run, which is 5-10 seconds too many:

import scriptcontext as sc
np = sc.sticky['numpy']

np.array(range(10000))

[pilcru]

The solution is to load the list first to the remote, then create the array from the remote list, to avoid numpy reading the items of the list one by one through the connection:

import scriptcontext as sc
np = sc.sticky['numpy']

r_range = rpyc.utils.classic.deliver(rpy, range(10000))
np.array(r_range)

This supposes:

• That Unicode-only strings in IronPython make rpyc.utils.classic.deliver() fail if the server is Python 2 tomerfiliba-org/rpyc#251 is fixed, so use a version after tomerfiliba-org/rpyc@e97f860
• That the GH components have a reference to the connection in rpy.