EVA-3423 - Fixes from tests

[tcezard]

This is a relatively large PR that refactor several things but should not change any of the core functionality

• Put all the scripts that are meant to be executable in a bin directory
• Rename the main script from eva_sub_cli.py because python confused it with the module eva-sub-cli.py see here
• Remove some command line argument that should go in a config file
• Make sample_checker.py take full path rather than relative path to the VCF to make sure they can be found
• Make sure the config and report are created in the submission directory

There are still issues post validation

[apriltuesday]

Is there a reason not to pass in vcf_name directly rather than re-compute in each method? Same for the assembly check files below.

[tcezard]

No reason in particular except that I didn't think the function should assume the string would only contain the name. I can pass the vcf_name and remove the extra call.

[apriltuesday]

I found this method confusing, maybe could use a docstring... I guess it's checking concordance between the VCF files provided in the metadata (files_per_analysis) and the ones passed on the command-line (vcf_files), the latter of which gives the full path... if this is true it feels like the resolving of location is kind of incidental now.

As an aside, I think I understand better the complexity you were talking about, it kind of feels like we should only receive metadata on the command line (including the files/analysis association) and use that as our sole source of truth. Then the metadata is more of a self-contained, complete description of the project, and there's less room for error (both user error and programmer error).