Replace nanofilt with filtlong to reduce the number of bad small reads

[Thomieh73]

This to implement a method that is more tuneable with respect to keeping the best reads in the dataset.

[Thomieh73]

Or should I replace it with Rasusa: https://github.com/mbhall88/rasusa

[Thomieh73]

https://joss.theoj.org/papers/10.21105/joss.03941

[Thomieh73]

See also this paper https://www.microbiologyresearch.org/content/journal/mgen/10.1099/mgen.0.000294

An alternative strategy deployed to reduce the computational burden of hybrid assembly was to randomly subsample long reads until a certain expected coverage was reached. [Table 1](https://www.microbiologyresearch.org/content/journal/mgen/10.1099/mgen.0.000294#T1) shows that this strategy was preferable to read correction and filtration: it did not reduce assembly completeness but did reduce computation time.

[Thomieh73]

Replace Nanofilt with Chopper.

Filtlong and Rasusa should be used differently, since there goal is really to reduce the complexity of the dataset.

[Thomieh73]

https://github.com/wdecoster/chopper

[Thomieh73]

See also issue #36