[RLlib] Fixed a bug with kl divergence calculation of torch.Dirichlet distribution within RLlib

[kouroshHakha]

Why are these changes needed?

context:
https://discuss.ray.io/t/passing-custom-action-dist/10111

The fix, basically lets the class inherit the default behavior of the parent for computing the kl_divergence which is the correct way of doing it.
self.dist.kl_divergence is not a valid api, instead we should use torch.distributions.kl.kl_divergence(self.dist, other.dist)

Related issue number

Checks

• I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/master/.
  • I've added any new APIs to the API Reference. For example, if I added a
    method in Tune, I've added it in doc/source/tune/api/ under the
    corresponding .rst file.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested :(

[avnishn]

is there a corresponding test?

Or do you need to merge as is?

[kouroshHakha]

is there a corresponding test?

Or do you need to merge as is?

I didn't want to spend more time on writing a test for this. The current test suite that we have for action distribution is not written elegantly and is not readily applicable to this. So it will be a time sink. IMO, we need to instead put that effort in testing our new distribution suite properly.