design

high level process
1. plates are registered in database - det_plate name, poolset id
2. plates are registered in queue_manager for PRISM analysis
3. yank identifies finished plate, notifies queue_manager
4. queue_manager puts plate in assemble queue
5. assemblify_queue checks queue, gets plate to work on, calls assemblify for that plate
      provides:
          id of new plate
          other plates that are part of the PRISM replicate and their yank status
          project ID
6. assemblify checks queue, gets plate to work on, prepares information for assemble
7. assemble generates gct
8. assemblify reports finish to queue_manager



assemblify config file
  contains prod and dev sections

  location of cell_id table - has meta data for cells

  database hostname - for use in mapping between filenames and plate meta data

inputs
  poolset information file - for each poolset, indicate the mapping between the
      analyte and the cell id

  location of csv files for each poolset

database
  extend lims_plate_orm to have new fields for PRISM


process
1.  load file mapping poolset to pools
2.  load file mapping cells to pools (and poolsets)
    look for consistency between these
3.  load file containing mapping between csv file and poolset