fflas-ffpack: usage of Modular<double> and getMaxModulus()

[jakobkroeker]

to get F_2 work using fflas-ffpack we switched to Modular<double>. But for Modular<double> the threshhold bound for delayed modular reduction in matrix-matrix computations was incorrectly computed, see https://groups.google.com/d/msg/ffpack-devel/aAEMSKIFT5I/9J5-exI7Jr0J

todo:

• check, if we use a version with the corresponding bugfix, and if yes, remove the hack in ARingZZpFFPACK::getMaxModulus()
• recall and document the issues with F_2 and ModularBalanced<double>

offtopic remark: since the ffpack authors use ModularBalanced<double> by default, Modular<double> was not well-tested.