Restructured pipeline

This release if the first since complete restructuring of the pipeline:

• A Singularity container is now defined for each rule, instead of having a big and hard to maintain container as previously.
• Pipeline preprocessing and formatting for the different classifiers is now in a dedicated worfklow.
• Original RDP is now implemented, with pretraining of the classifier during DBprocessing and optional rules to validate its performance.
• Reference databases were removed of this git repository (were too big). Their location (path) and name (directory at this path) must now be defined in config.
• Cutadapt settings were changed to filter out sequences without a primer at the expected position. This was implemented to avoid the inclusion of sequences in the wrong direction which would mess with phylogeny.

Traceability is now improved by :

• The recovery of the current commit and user in logs, including for DB processing.
• Hash generated during DB processing

Future improvements:

• Check for reference DB hash before tax assignment.
• Set the Shinyapp in a container.
• Generate a report to QC the whole process.
• Document with Read the Doc.
• Implement Travis to validate releases.