Improve species identification reliability on edge cases or poor quality inputs

[kbessonov1984]

• the RefSeq sketch is stale and needs to be updated more frequently to match with the NCBI RefSeq database sketch.
• Also in species prediction module implement species prediction threshold via min MASH score check in cases when all top hits are equal to MASH p-value of 1 (no certainty) due to poor WGS data quality input (e.g. truncated library). In this case, return species "-" instead of some erroneous prediction