Refactoring

[HadrienG]

Issue to track the progress on the Roadmap item "Refactor to avoid duplicated code when possible"

Some functions are duplicated at the moment, and the bam parsing and model creation is quite messy.

• There are duplicated functions in the error model classes. The indels and substitutions functions can probably be moved to the base ErrorModel class
• Ideally a bam file should only be read once to create the model .npz file. It is not the case now.
• Some functions are too long (notably in bam.py) and should be broken done into smaller functions. Possibly the bam.py file could even be split in 2 files (perhaps 1 general bam reading / writing and 1 where the heavy lifting / model creation happens)

[HadrienG]

closed with ba7ef2d