Better handling of several special cases for TransitivePaths.

[floriankramer]

This pr attempts to improve the speed of TransitivePath operations by limiting the number of source vertices. To this end:

• TransitivePath operations with a fixed right end now invert the input edges and run a dfs from the single right node
• A transitive path operation can replace a join between a transitive path and another operation to reduce the amount of tree explorations (dfs) that need to be run

[floriankramer]

The query

PREFIX p: <http://www.wikidata.org/prop/> 
PREFIX psn: <http://www.wikidata.org/prop/statement/value-normalized/> 
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> 
PREFIX wd: <http://www.wikidata.org/entity/> 
PREFIX wdt: <http://www.wikidata.org/prop/direct/> 
PREFIX wikibase: <http://wikiba.se/ontology#>
SELECT ?person ?r WHERE {
  ?person (<http://www.wikidata.org/prop/direct/P166>)/((^<http://www.wikidata.org/prop/direct/P31>)?) ?r
}

Is incredibly slow, as the inverse of P31 has a lot of edges for every node on the left side. To handle that case better I switched from vectors to HashSets for the storage of a nodes edges in the latest commit.

[niklas88]

This looks great, I found two copy and paste fix omissions but other than that the code looks solid as always.

[niklas88]

s/left/right/

[niklas88]

s/left/right