Version less memory intensive and better MT handling + merge bams w/o vcf

[clementhelsens]

This Pull Request contains:

• reduction of memory consumption
• multi threading based on splitting the positions and number of input files
• option --merge when no input vcf is use to still use the synth polling
• option --shuffle to shuffle the input positions (makes each thread length more uniform)
• option --test to run only a subset of positions

[clementhelsens]

hello @huangyh09 , I pushed a last change this morning with a better naming convention. For your information, this updated version is being used in the coming release of this paper.
Happy to discuss the content of this PR on GitHub or outside!

[huangyh09]

Thanks @clementhelsens for the efforts/contributions and make the pull request. I'll review the codes and merge them in a week or so.