[RLlib] Better error message for partial space mismatch (Dict|Tuple) between env-returned values and given action/obs space.

[sven1977]

Signed-off-by: sven1977 [email protected]

Create a better error message for when parts of the observations/actions (Dict or Tuple space) do not match the actual space.
Right now, the user is left with a cryptic printout of a very large dict of numpy arrays that supposedly do not match a very large and complex Dict/Tuple space.
The PR adds the exact key/index-path, the single offending value and sub-space, as well as the dtypes to the error message, such that finding the problem becomes much easier.

Why are these changes needed?

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 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 :(

[ArturNiederfahrenhorst]

lgtm!