Improve parsing robustness

## Description

The [LOD Laundromat](http://lodlaundromat.org/) has shown that
* Online several RDF datasets contain errors
* They still can be parsed with robust parsers

The goal of this issue is to adapt the robustness of the LOD Laundromat to Squirrel.

## Solution
* The LOD Laundromat has been programmed in Prolog so we can not use the code "directly" in our Java crawler
* It might be possible to have a step between the fetcher and the analyzer which makes sure that the RDF is valid, i.e., instead of `Fetcher -> File A -> Analyzer` we could have `Fetcher -> File A -> RDF Checker -> File B -> Analyzer` where the RDF Checker could be a Prolog program using the LOD Laundromat code.

The issue names no files or tests; begin by tracing the existing Fetcher -> File A -> Analyzer flow in the Java crawler. Compare it with the proposed RDF Checker step and determine how malformed RDF should be handled before the Analyzer; the issue does not define a concrete completion test or implementation boundary.

Written by the indexing model from the issue text.