fix implementation of mu-e conversion

[peterstangl]

This PR corrects several errors in the implementation of mu-e conversion, in particular

• A wrong normalization was used for gRV, gLV, gRS, and gLS as pointed out by @xmarcano in issue #126
• No distinction was made between the proton and neutron versions of GuS, GdS, and GsS as also pointed out in issue #126

In addition, while fixing the above issues, I found also others:

• AL and AR were proportional to a conjugated and an unconjugated flavio Wilson coefficient, respectively, while it should be the other way around
• gRV, gLV, gRS, and gLS were expressed in terms of the wrong flavio Wilson coefficients (in addition to the wrong normalization)

All of the above issues should be fixed by this PR. @xmarcano and @Broko55, could you maybe take a look and confirm that my modifications are indeed correct? I used the expressions from https://arxiv.org/pdf/hep-ph/0203110.pdf and for the flavio Wilson coefficients I used https://wcxf.github.io/assets/pdf/WET.flavio.pdf.

[coveralls]

Coverage increased (+0.002%) to 93.872% when pulling ca07bd5ec2d26ec775186138e066bccae1ed5ba0 on peterstangl:mue_conversion into d4f372bb43a69b944b1d8fa1b34870beab89a551 on flav-io:master.

[peterstangl]

Thank you @xmarcano and @Broko55 for approving the changes!