Errors models in 0.2.1 don't work

[HadrienG]

The SRA reads taken for building the new error models were not all the same length (adapter trimming, possible quality too) and it caused issues during the model generation 😱

Will rebuild asap with either modified SRA reads or my coral metagenome reads (need to assemble and map back to assembly then)

In the meantime, I've rolled back with the old error models (named HiSeq2500 and MiSeq) and push to 0.2.2 (also on Pypi)

[HadrienG]

Closed with the last release and the introduction of updated error models