Historically (from what I was told) these were not in the classpath because there existed other version of these tools that users wanted to run and hence did not want the hadoop tools to be in the classpath by default. Is that still true today?
distcp is the big one. We definitely run a custom version as do many other people. Additionally, at least in 0.20.2, there isn't much of a reason to include them in the CLASSPATH:
- har creation
- logalyzer (which I don't even know what that is)
all seem pretty non-fundamental to me.
Looking at 1.x, I still see a lot of non-fundamental bits that most people just really don't need.