use CMake FetchContent instead of git submodules

[BenWibking]

Describe the proposal
Git submodules are painful to use. An alternative is to use CMake's FetchContent feature. That requires specifying the git commit that is needed in the CMakeLists.txt file instead:

FetchContent_Declare(
  googletest
  GIT_REPOSITORY https://github.com/google/googletest.git
  GIT_TAG        703bd9caab50b139428cea1aaff9974ebee5742e # release-1.10.0
)
FetchContent_Declare(
  myCompanyIcons
  URL      https://intranet.mycompany.com/assets/iconset_1.12.tar.gz
  URL_HASH MD5=5588a7b18261c20068beabfb4f530b87
)

FetchContent_MakeAvailable(googletest myCompanyIcons)

This requires that the machine on which you run the cmake .. configure step has internet access, which is not the case on compute nodes on many clusters. However, the configure step could be run on the login nodes, and then the build step (only) can be run on the compute nodes.

The major disadvantage is that GitHub's Dependabot will NOT auto-update the GIT_TAG (shown above) to the latest version on a periodic basis (but see this issue: dependabot/dependabot-core#7451). Currently, we depend on Dependabot to do this weekly.

Describe alternatives you've considered
Keep using git submodules.

Additional context
It is not possible to edit the source code of the dependency for quick testing purposes, since CMake always fetches and uses the version specified in the CMakeLists.txt file:
https://cmake.org/pipermail/cmake/2019-June/069709.html