The CI tests for submodule master are not doing what we expect

[thunderbiscuit]

I think they are currently running the committed hash of the submodule rather than testing the new commit pushed on master on bdk-ffi. This is due to the build scripts explicitly performing git submodule update --init. We should ensure we have the option of not doing this, so that the build script can actually build using a submodule checked out at a different commit.

In bdk-jvm, this is what we have in the scripts:

printf "\nSubmodule check...\n"
if [[ "$1" != "--skip-submodule-update" ]]; then
  git submodule update --init
  printf "Submodule is checked out at commit: $(git submodule status)\n\n"
else
  printf "Skipping submodule update, using local changes.\n"
  printf "Submodule is checked out at commit: $(git submodule status)\n\n"
fi