work in progress for version 2.0; fixed test; still todo : merge tests with those from version 1.9.6

bugfix: when the number of k-mer is provided in command line by the user, it was implicitely converted to unsigned int instead of unsigned long which led to a wrong value for large numbers of k-mets.

added option -f to allow thge users to specify what maximum value of collision probability they want in the CMS.

re-used what I had implemented for the -g option to warn the user that there is not enough memory on the machine to process all reads with the given or default low filter.

 bugfix when PE treated as separated, undef filter considered them as single. bugfix when minimum quality score for nucleotides is 0, nucleotides with score 0 were considered as errors. -q option now takes  argumens (see News file and doc)

version 1.7: users now have the possibility to specify a minimum number of k-mer when processing PE separately.

changes: input filenames can now be provided as argument and output filenames can be generated by ROCK.