update filter_n_probes rule to also drop seq candidates with '-'

[conorcamplisson]

Any homology sequence candidates that contain '-' characters will break the calc_prob and max_kmer Snakemake rules. We suspect these are introduced by sam2pairwise when >60 nt homology lengths are allowed (i.e. by modifying the blockparse_min_length and blockparse_max_length parameters inside config.yml) via invocation of its 'Hard Clipping' mode. The existing filter_n_probes Snakemake rule should be updated to also drop any sequence candidates that contain '-'.