Important compiler workarounds and code fixes for NVIDIA

Current versions

Package	Version
gFTL	v1.8.3
gFTL-shared	v1.5.1
fArgParse	v1.4.2
pFUnit	v4.6.3
yaFyaml	v1.0.7
pFlogger	v1.9.3

---

Fixed

• Update gFTL to v1.8.3
  • Fixes for GNU Make builds
  • Missing KIND= on size() procedure. Gave incorrect response for large containers. Only affects V1 containers.
• Update gFTL-shared to v1.5.1
  • Fixes for GNU Make builds
• Update fArgParse to v1.4.2
  • Fixes for GNU Make builds
• Update pFUnit to v4.6.3
  • Fix for compilers that do not support 128 bit reals
  • Fixed build_submodule for old git versions
  • Fixed for use with FetchContent
  • Fix CMake logic in add_pfunit_ctest.cmake for MPIEXEC_EXECUTABLE. Problem not exposed by common MPI flavors which use mpirun
  • Fix GitHub CI workflow by pinning to CMake 3.24.3
  • Fixes for GNU Make builds
  • Update fArgParse submodule to v1.4.2
  • Fix pFUnitParser.py in cases where there is no module name
• Update yaFyaml to v1.0.7
  • Workaround for NVIDIA compiler
  • Fixes for GNU Make builds
  • Workarounds for Intel 2021.7
  • Out of date examples have been updated
• Update pFlogger to v1.9.3
  • Fixes for GNU Make builds
  • Package was incorrectly assuming that all extant compilers support 128 bit reals. Now a check is performed to optionally include support for 128 bit reals.
  • Fix Parser() declaration for Intel 2021.7