Add support for processing tabular data from select queries

[blcham]

SPARQL is inefficient to process tabular data respecting an order of that data.

---

Use cases

Use case 1: inference of has-next-trail-point:

• slow/unusable query: https://github.com/kbss-cvut/24lkpr-distribution/issues/21#issuecomment-2554442413

PREFIX : <http://onto.fel.cvut.cz/ontologies/safety-performance-kpis/>

INSERT {
    GRAPH :has-next-trail-point {
    	?timePoint :has-next-trail-point ?nextTimePoint    
    }    
}
WHERE { 
    ?trail a :trail .
    
    ?trail :has-point ?timePoint .
    ?timePoint :date ?date .
    
    ?trail :has-point ?nextTimePoint . 
    ?nextTimePoint :date ?nextDate .
    
    FILTER(?nextDate > ?date) 
    
    FILTER NOT EXISTS {
      ?trail :has-point ?nextTimePointNE . 
      ?nextTimePointNE :date ?nextDateNE .
      
      FILTER(?nextDateNE > ?date)
      FILTER(?nextDate > ?nextDateNE)
    }
}

• solution:
  • 1. create ordered tabular data (see SELECT query)
  • 2. use order to compute final inferences (see not existing predicates with suffix __previous())

PREFIX : <http://onto.fel.cvut.cz/ontologies/safety-performance-kpis/>

CONSTRUCT {
  ?timePoint__previous :has-next-trail-point ?timePoint .
} WHERE {
  SELECT ?trail ?trailId ?timePoint ?date {
    ?trail a :trail .  
    ?trail :id ?trailId .
    
    ?trail :has-point ?timePoint .
    ?timePoint :date ?date .
  } ORDER BY ?trailId ?date 
  
  FILTER(?trial__previous = ?trial)   # note that ?trial__previous does not exist but could be provided from the SELECT query above
}

A/C:

• implement extension of apply-construct-with-chunked-values
• implement extension of apply-construct-with-chunked-values-and-scrollable-cursor
• implement extension of tabular module