Change handling of spike ins

[rmcolq]

Discuss. But likely we want scylla to reevaluate proportions for the JSON passed to kraken to exclude things which are spike

[rmcolq]

Steps:

1. Update initial processing for spike ins to remove all spike reads via mapping -> have some info about quantity of spike reads to add to onyx
2. After filtering spike reads, we run classification -> the percentages in onyx are percentages excluding spike