Census tweaks

[WPettersson]

Outputs more useful information to stdout

Instead of just facepairing.str(), tricensus will now output a comma-separated list for each face pairing graph analysed. The values printed are:

index, graph, running time, count

defined as follows.

• index is either a line number of the input pairs list, or for generated pairs index tells us when in the sequence the graph was generated
• graph is the text representation of the pairing graph
• running time is the running time taken to process said graph
• count is the number of relevant graphs returned

Only parse every x'th line of input

Note: this feature is only functional if graphs are passed via stdin.

tricensus has two new parameters, -x and -y. When appropriately assigned integers, tricensus will only analyse a graph of index i if i ≅ x mod y. This makes it trivial to parallelise tricensus on clusters without having to pre-plan the graph splits. Note that this definitely won't guarantee equal running time, but any jobs that terminate early will return resources to the cluster queue.

Allow %d in output file name

If the output file name given to tricensus contains %d it will be replaced with the index of the given graph (so a new file is created for each graph). This can be used to replicate the output style used by tricensus-mpi.

[WPettersson]

I've got this in a personal branch forked from a pre-renaming master. I'll re-base it off current master once we finish re-naming and push it then, to make it easier to follow.

[WPettersson]

Should add a flag to enable the CSV output, so as to not break existing code.