Addresses the first test failure on CIT mentioned in #12 (closed). The same workaround is already present in core.py, not sure why on other systems I hadn't seen the import problem triggered from within helper_functions.py like it now does.
Note other files still have the plain import, but in all scenarios I tried, fixing these two seems to be sufficient. I'd assume enforcing agg once per session is sufficient, and only in some special circumstances are the helper_functions called before the import from core, whereas grid_based_searches and mcmc_based_searches always have had the import from core before they call matplotlib themselves.