[MT-TREE] Support prediction cache and model slicing. (#8968)

- Fix prediction range.
- Support prediction cache in mt-hist.
- Support model slicing.
- Make the booster a Python iterable by defining `__iter__`.
- Cleanup removed/deprecated parameters.
- A new field in the output model `iteration_indptr` for pointing to the ranges of trees for each iteration.

tests/cpp/gbm/test_gbtree.cc

@@ -505,7 +505,7 @@ TEST(GBTree, PredictRange) {
     auto h_out_predt_full = out_predt->HostVector();
 
     ASSERT_TRUE(std::equal(h_out_predt.begin(), h_out_predt.end(), h_out_predt_full.begin()));
-
+    // Out of range.
     ASSERT_THROW(learner->InplacePredict(x, PredictionType::kValue,
                                          std::numeric_limits<float>::quiet_NaN(), &out_predt, 0, 3),
                  dmlc::Error);