refactor sim code

[knaaptime]

• use dask in acs processing
  • use ftp as default when downloading demographic profile
• fix iteritems error
• rm proplot dependency
• update code for drawing states from spatial Markov
• rm outdated deps and test on newer python
• update workflows

[knaaptime]

@sjsrey @weikang9009 @ljwolf this PR refactors the code for drawing states from the SMC to generalize and make more transparent. If you're into it, it might be an opportune time to move some over into giddy to address pysal/giddy#161

one or two pieces can still be parallelized (left comments where) and the seeding still isnt working perfectly. I think the mechanics are right, but i cant generate reproducible results reliably. Might have something to with the tolerance/rounding coming out of the SMC

[knaaptime]

also addresses #368