Double deletion (#624)

* intermediate step towards double deletion

* make tests pass

* very minimalistic
* only single processing

* add mini script to run double_reaction_deletion

* working prototype of double reaction deletion

* do not track benchmark script

* feat: refactor single and double deletions

* fix: fba tests

* fix: do not remove zero flux reactions; small optimisations

* fix: moma

* style: pep8

* fix: do not use legacy solvers to test new deletion functionality

* fix: no context manager for python2.7

* fix: assign the attribute to the function during process initialization

* fix: use worker class

* fix: use global variables to pass model to processes

* refactor: do not use global variable with one process

* refactor: construct biomass; extra imports

* fix: do not import moma if there is no scipy

* refactor: absolute imports