This commit introduces a new helper facility,
data::InterRegFlowMap
that knows how to interpret region IDs. It is a simple wrapper
around a common CSR representation of (source, destination) region
ID pairs with the additional feature that the element type of the
connection is a linear sequence values of arithmetic type. We've
initially elected to make this sequence out of 'float' in order to
conserve memory since we expect to do little arithmetic to the
sequence itself outside of compression and parallel communication.
Client code will be of two primary categories
1. Producing side which inserts flow rates from individual
connections, compresses the result when done, and communicates
the results to one or more parallel I/O ranks
2. Consuming side, typically on an I/O rank, which merges the
contributions from multiple source ranks and extracts values
pertaining inter-region flows for purpose of outputting time
series data to a summary file.
The first case will create and populate 'FlowRates' and add these
into the container using the 'addConnection' function and finally
call 'compress' when complete. Consumers will then use member
function 'getInterRegFlows' to extract a signed view of flow rate
values associated to a particular region ID pair (nullopt if no such
pair is registered).
