Improved multi-node multi-GPU random forests. (#4238)

* Improved multi-node multi-GPU random forests.

- removed rabit::Broadcast() from each invocation of column sampling
- instead, syncing the PRNG seed when a ColumnSampler() object is constructed
- this makes non-trivial column sampling significantly faster in the distributed case
- refactored distributed GPU tests
- added distributed random forests tests

tests/ci_build/test_mgpu.sh

@@ -7,5 +7,5 @@ cd ..
 pytest -v -s --fulltrace -m "(not slow) and mgpu" tests/python-gpu
 ./testxgboost --gtest_filter=*.MGPU_*
 
-cd tests/distributed-gpu
-./runtests-gpu.sh
+cd tests/distributed
+./runtests-gpu.sh