Sub-requests for large molecules causing issue

[lambdacasserole]

If we search chlorophyll for example we get a very long SMILES string:

[Mg++].CCc1c(C)c2[n-]c1cc3nc4c([C@@H](C(=O)OC)C(=O)c4c3C)c5nc(cc6[n-]c(c2)c(C=C)c6C)[C@@H](C)[C@@H]5CCC(=O)OC/C=C(C)/CCCC(C)CCCC(C)CCCC(C)C

Which becomes outright monstrous when URL-encoded:

%5BMg%2B%2B%5D.CCc1c%28C%29c2%5Bn-%5Dc1cc3nc4c%28%5BC%40%40H%5D%28C%28%3DO%29OC%29C%28%3DO%29c4c3C%29c5nc%28cc6%5Bn-%5Dc%28c2%29c%28C%3DC%29c6C%29%5BC%40%40H%5D%28C%29%5BC%40%40H%5D5CCC%28%3DO%29OC%2FC%3DC%28C%29%2FCCCC%28C%29CCCC%28C%29CCCC%28C%29C

This breaks sub-requests in Silex as far as I can tell. The only real solution is to avoid passing these strings from action to action and deal with them in the same action they're generated in. This is going to create some redundant code for the time being.

[lambdacasserole]

Fixed by #20.