Add fast update function

[johnlees]

Fixes #321

[nickjcroucher]

Nice - took me a while to understand the cluster merging, but looks good - sorry for not spotting this needed a review earlier. Is it worth addding a test for the fast pruning mode? Also based on the current CI tests it looks like the extractReferences call in __main__ is missing a merged_queries argument.

[johnlees]

python /home/jlees/installs/PopPUNK/poppunk_assign-runner.py --db Salmonella_progressive/update_0 --previous-clustering Salmonella_core_threshold --distances sal_sketch40k/sal_sketch40k.dists --query Salmonella_query_files/salmonella_split.txtab --model-dir Salmonella_core_threshold --threads 16 --output Salmonella_progressive/update_1_fast --update-db fast

Graph-tools OpenMP parallelisation enabled: with 40 threads
Looking for existing sketches in Salmonella_progressive/update_1/update_1.h5
Loading previously refined model
Completed model loading
48180 refs 31935 queries

WARNING: versions of input databases sketches are different, results may not be compatible
Calculating distances using 40 thread(s)
Progress (CPU): 100.0%
Loading network from Salmonella_core_threshold/Salmonella_core_threshold_graph.gt
Network loaded: 48180 samples
Loading previous cluster assignments from Salmonella_core_threshold/Salmonella_core_threshold_clusters.csv
1538628300 assignments 48180 refs 31935 queries

109549943 tuples

Calculating all query-query distances
Calculating random match chances using Monte Carlo
Calculating distances using 40 thread(s)
Progress (CPU): 100.0%

509906145 assignments 80115 refs 80115 queries

37646978 tuples

[johnlees]

core_threshold: 1333 refs
update_0: 1746 refs

[johnlees]

Most recent run:

Loading previously refined model
Completed model loading
1333 refs 31935 queries

WARNING: versions of input databases sketches are different, results may not be compatible
Calculating distances using 16 thread(s)
Progress (CPU): 100.0%
Loading network from Salmonella_core_threshold/Salmonella_core_threshold.refs_graph.gt
Network loaded: 1333 samples
Loading previous cluster assignments from Salmonella_core_threshold/Salmonella_core_threshold_clusters.csv
42569355 assignments 1333 refs 31935 queries

Calculating all query-query distances
Calculating random match chances using Monte Carlo
Calculating distances using 16 thread(s)
Progress (CPU): 100.0%
509906145 assignments 33268 refs 33268 queries

Clusters 16,995 have merged into 16_995
Clusters 82,168 have merged into 82_168
Clusters 136,550 have merged into 136_550
Clusters 231,563 have merged into 231_563
Clusters 243,466 have merged into 243_466
Clusters 320,1052 have merged into 320_1052
Clusters 359,666 have merged into 359_666
Clusters 348,464 have merged into 348_464
Clusters 329,756 have merged into 329_756
Clusters 428,696 have merged into 428_696
Clusters 620,887 have merged into 620_887
Updating reference database to Salmonella_progressive/update_1_fast
Updating random match chances
Calculating random match chances using Monte Carlo
Saving model and network
Finding references (fast)
Running quick reference picking

then errors:

multiprocessing.pool.MaybeEncodingError: Error sending result:
...
Reason: 'RuntimeError('Pickling of "graph_tool.libgraph_tool_core.Vertex" instances is not enabled (http://www.boost.org/libs/python/doc/v2/pickle.html)')'

note some <invalid Vertex object at 0x7f7986d6e440> could be causing this?