[EM] Support mmap backed ellpack. (#10602)

- Support resource view in ellpack.
- Define the CUDA version of MMAP resource.
- Define the CUDA version of malloc resource.
- Refactor cuda runtime API wrappers, and add memory access related wrappers.
- gather windows macros into a single header.

src/gbm/gblinear.cc

@@ -10,12 +10,12 @@
 
 #include <algorithm>
 #include <numeric>
-#include <sstream>
 #include <string>
 #include <vector>
 
 #include "../common/common.h"
-#include "../common/error_msg.h"  // NoCategorical, DeprecatedFunc
+#include "../common/cuda_rt_utils.h"  // for AllVisibleGPUs
+#include "../common/error_msg.h"      // NoCategorical, DeprecatedFunc
 #include "../common/threading_utils.h"
 #include "../common/timer.h"
 #include "gblinear_model.h"

src/gbm/gbtree.cc

@@ -1,5 +1,5 @@
 /**
- * Copyright 2014-2023 by Contributors
+ * Copyright 2014-2024, XGBoost Contributors
  * \file gbtree.cc
  * \brief gradient boosted tree implementation.
  * \author Tianqi Chen
@@ -10,14 +10,14 @@
 #include <dmlc/parameter.h>
 
 #include <algorithm>  // for equal
-#include <cinttypes>  // for uint32_t
-#include <limits>
+#include <cstdint>    // for uint32_t
 #include <memory>
 #include <string>
 #include <utility>
 #include <vector>
 
 #include "../common/common.h"
+#include "../common/cuda_rt_utils.h"  // for AllVisibleGPUs
 #include "../common/error_msg.h"  // for UnknownDevice, WarnOldSerialization, InplacePredictProxy
 #include "../common/random.h"
 #include "../common/threading_utils.h"
@@ -347,7 +347,7 @@ void GBTree::LoadConfig(Json const& in) {
   // This would cause all trees to be pushed to trees_to_update
   // e.g. updating a model, then saving and loading it would result in an empty model
   tparam_.process_type = TreeProcessType::kDefault;
-  std::int32_t const n_gpus = xgboost::common::AllVisibleGPUs();
+  std::int32_t const n_gpus = common::AllVisibleGPUs();
 
   auto msg = StringView{
       R"(