Fix phaseSpaceIntegral and small stuff

[suhlatwork]

Fix the values calculated by the phaseSpaceIntegral class. The volume of the integration was taken into account twice, once in the normalization of the S_U_CHUNG weight, and once explicitly in the calculation this class. Now only use the one from the normalization.

Also issue #91 is partially fixed. If the two Lorentz-vectors of beam and X are equal and X is at rest (a small variation is allowed), the boost matrix, which is basically the unit matrix except for the small allowed variation, is returned. The boost is also performed if X is not at rest, however, then a warning is printed.

The effects of those two corrections are depicted here:

Also fix a couple of smaller stuff:

• the copy and assignment constructors of the waveDescription should only parse the key file content, if they were successfully parse before, otherwise it is not possible to copy an empty waveDescription object without an error message.
• fix to compile with -Wextra
• fix normalization of amplitudes in view of zero entries in the normalization integral, zero entries might occur when using the flatRange mass-dependence, however, then integral and amplitude should be zero. So in case of zero amplitude and integral do nothing (the amplitude remains zero (but does no longer become NaN spoiling everything)), in case of zero integral and non-zero amplitude print an error and exit because this should neverhappen, and in the case of non-zero integral do the normalization. (Note that this was rather poorly fixed in case of the binary amplitudes already before.)

[legordian]

Thank you very much for these fixes! Merged.