Release mode doesn't run std::isfinite, isnan, isinf etc.

[danieldeidda]

When installing STIR in release mode the cxx tag uses fast math however this does not allow to use some std functions like isfininit, isnan, isinf etc.
when installing with and withouth fast math math the stir_timing outputs are as follow:

so mostly fast math does make thing slightly faster but not sure that it would make any significant difference in a reconstruction

[KrisThielemans]

Some more info on support is at https://en.cppreference.com/w/cpp/numeric/math/isnan

GCC and Clang support a -ffinite-math option (additionally implied by -ffast-math), which allows the respective compiler to assume the nonexistence of special IEEE-754 floating point values such as NaN, infinity, or negative zero. In other words, std::isnan is assumed to always return false under this option.

Suggestion is to

• add a CMake variable ENABLE_FASTMATH defaulting to ON, which creates a preprocessor variable (via STIRConfig.h.in) STIR_WITH_fastmath
• any utility/function that uses std::isfinite etc should throw an error #ifdef STIR_WITH_fastmath