Bugfix pythonic cell construction

[stonier]

This does the following:

• Allows a cell name override to be passed in **args (just like c++)

This was conspicuously causing an error if you tried - the c++ api for cell construction didn't take an arg. I noticed the python test was dodging this bullet by not passing along *args at all. I couldn't work out how to get this to work with the c++ constructor call, but managed to work around by delaying name setting till after the __init__.

• Removes the configure step from construction

This syncs it with c++ behaviour and opens up the possibility of getting it working with directed configuration (#260).

• Adds input and output tendrils to the configure step to sync with c++ api
• Bugfixes self.__dict__.hasget (pretty sure dict has no hasget method ;))

While messing around I had this getting thrown, but I think it was just hiding because it doesn't get accessed so often. Certainly not in the tests.

• Updated the python module test and made sure it is testing for naming as well.

[stonier]

Just re-reading this issue and obviously looks like my brain was deep into the problem. Just to put it more simply this is what you can do for a c++ cell in an ecto script:

i1 = Increment('Inc 1', start=18)

which sets the name of the cell to 'Inc 1'. But, if the cell is a python cell, this causes problems. This pull request fixes that along with a couple of other minor things.

[vrabaud]

I was out for some time, sorry for the delay. I guess BlackBox is what is used mostly. Do you want to be maintainer for the ecto stuff ? I'll give you rights if you want: this way you can merge any trivial fix.

[stonier]

Do you want to be maintainer for the ecto stuff ? I'll give you rights if you want: this way you can merge any trivial fix.

Sure. Looks like we'll be using ecto for the foreseeable future.

[stonier]

I started using these python cells for directed configuration. Some things are just far easier in python and configuration isn't a step that needs speed.