added SIMD version of classic gamma index (Wendling method isn't vectorized - there were two unsuccessful attempts) with xsimd library. To select this version set cmake option GAMMA_VERSION to SIMD.
added cmake option SIMD_EXTENSION for choosing which SIMD extension you want to use (it can also be used in non-SIMD version of gamma index for auto-vectorization)
added SIMD version to Github Actions CI
added cmake option ENABLE_FMA to set -mfma compilation option (only on compilers that support it). It gives better performance, but slightly different results
sped up classic gamma index by precalculating coordinates
sped up Wendling gamma index by simplifying its interpolation calculations
List of changes:
GAMMA_VERSIONtoSIMD.SIMD_EXTENSIONfor choosing which SIMD extension you want to use (it can also be used in non-SIMD version of gamma index for auto-vectorization)ENABLE_FMAto set-mfmacompilation option (only on compilers that support it). It gives better performance, but slightly different results