Just try this little experiment: disable the bash-completions package, and see how much more quickly you can open a new shell. While it might seem miniscule on your multicore multigigabyte desktop machine, on a NAS with limited resources, it can make a significant difference. And when you are scripting something (say, a nightly rsync backup), do you really need to have tab completion for git? This also applies to things you want to run once (on login) versus every time you open a new shell.
Just because you can't think of a use for something doesn't mean no one else can. Usually there is a reason for a feature, and the nice part is that nine times out of ten, when I'm looking for a way to do something I hadn't thought of before, it's already there in UNIX/Linux.
Let me put it another way: don't try to outthink the users' intentions- leave it to me and my scripts to decide what to run and when. My script should make the decision whether it's appropriate to load certain things. Passing --interactive, --login, --remote, --wtfe to the shell should cause nothing more than a flag to be set for my script to query. Then my script can make the decision what to load and when.
I do understand the original intent. I believe it was the wrong was to go about it.