neuron parameter optimization

[emuller]

Add a module for neuron parameter optimization.

Features:

• use ipython parallel python instances if provided by the user
• support heuristic search algos like genetic algorithms, and simulated annealing, as well as user defined search methods
• Support Multi-objective optimization
• semi-automation of neuron parameter fitting to user provided data.
• interaction with databases to scour neuron measurements for parameter statistics
• Example scripts for a straight-forward start for an entry to the single neuron modeling challenge:  http://lcn.epfl.ch/QuantNeuronMod/Challenge 2008.html

[mschmucker]

examples/parameter_search can use IPython for parallel parameter searching. A possible solution could build upon that.

Regarding search algorithms, we could use scipy.optimize.

[mschmucker]

Added a draft implementation in  rev 365, with an addendum in  rev 366. Only grid search is implemented right now, but I hope to be able to provide a non-trivial searcher soon :)