[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/tree/test_gpu_hist.cu

@@ -246,7 +246,7 @@ void UpdateTree(HostDeviceVector<GradientPair>* gpair, DMatrix* dmat,
   std::vector<HostDeviceVector<bst_node_t>> position(1);
   hist_maker.Update(&param, gpair, dmat, common::Span<HostDeviceVector<bst_node_t>>{position},
                     {tree});
-  auto cache = linalg::VectorView<float>{preds->DeviceSpan(), {preds->Size()}, 0};
+  auto cache = linalg::MakeTensorView(&ctx, preds->DeviceSpan(), preds->Size(), 1);
   hist_maker.UpdatePredictionCache(dmat, cache);
 }