deal with variable J

[ajrominger]

• replace localSpp.size() with J[step]
• update death method to know about step
• in each step, check for change in J
• if J is smaller, subsample local comm
• if J is bigger, run birthImm and speciation until get the right size
• calculate the contribution of each step toward generations during every step
• unit tests

[ajrominger]

as of 592fd03 we have code that attempts to do all tasks (except no unit tests) but I realized to truly change local community size we also need to update traits and distances (env and comp), which is a much bigger undertaking than can be done for now, and also current implementation requires incrementally adding and subtracting from local comm objects, which will be slower than if we can figure out how to buffer the objects with empty elements