This option was again source of confusion (running Vidjil without options was anyway non meaningful).
Now a V(D)J segmentation is run when a compatible germline is given (-D, -G, or -g),
and the correct D parameters are selected.

This was source of confusion, in particular in selecting or not D parameters.
Now the code is cleaner.

See 183ad875.

This code was duplicated in 60a8bf4a.

Simplification of the code.

Closes bug tested by 14b16e9c.

If -w is below the seed weight (hence the junction size), strange things may happen,
as for example a clone without representative.

(Now only the last option of each is really stored and indexed, see 1a05fe6a.)

The CMD_SEGMENT scenario now tries the germlines in the multigermline (until the read is segmented).

On the default file, Vidjil now returns some clones even with default options.

As KmerAffectAnalyser is only used with KmerAffect, we do not need to keep the templates.
The code should be easier to read for C++ newbies such as the author of this commit.

Accessing this method while the window is not label does not create an element in the map.

Moreover, now this is exported in .vidjil.

The usage said it was '-f', but the option was actually '-t'.
This option is for now not covered by the tests.

Reformulation of 6ec0f991, with a short notice on usage

* .data files are renamed to .vidjil
* by default output filenames have the same basename as the input file

Documentation and tests have been updated accordingly

This was hardcoded twice in the code.

This feature was not documented and not used.
The browser offers better normalization features.

Fixed thanks to Valgrind.

This may help the upcoming modularization of vidjil.cpp

(Soissons)