Run analysis in a separate Python process

[geoffxy]

For convenience, the analysis code runs in the same process as the Skyline server. However, the analysis code executes user code as well. Running user code in the main Skyline process is not really ideal:

• We need to mess around with sys.path to ensure the user's code can import their modules properly
• We need to prevent their modules from being cached in sys.modules so that we can re-run the analysis if they change their code

Security issues are less of a concern here because the user uses Skyline to run their own code on their own system.

If time permits, we should move the analysis into a subprocess launched by the Skyline server.