Can't pickle database

[rwest]

Most parallel processing frameworks require us to pickle things, like the database. Currently I get "TypeError: can't pickle instancemethod objects" when I try.

[jwallen]

It looks like we may just need to add a few __reduce__() methods to some of the database classes to help pickle understand how to handle them. I have managed to pickle the thermo database after doing this (it didn't work before), and am working on the kinetics and frequencies databases.

[jwallen]

Some timings from my test script (for the entire database):

Loaded database in 33.23 s
Pickled database in 11.54 s
Unpickled database in 24.81 s

[rwest]

Hmm - so storing it pickled wouldn't save us much.

@faribas this explains why our parallel implementation which requires pickling and unpickling the entire database for every edge species is kind of slow! (don't worry - we have a better idea...)

Thanks for fixing, Josh