fix(cuda.core): add 3 missing CUresult entries to frozen fallback table (#6843633)

cuda-bindings 13.4.2 exposes 106 CUresult members while cuda.core 1.2.1's
frozen fallback table (capped at CTK 13.1.1/13.3) only covers 103, so
test_frozen_driver_table_covers_all_curesult_members fails on every
Python version tested.

Add the 3 members introduced by cuda-bindings 13.4.1+:
- CUDA_ERROR_MULTICAST_RESOURCE_FULL (47)
- CUDA_ERROR_INSUFFICIENT_LOADER_VERSION (227)
- CUDA_ERROR_FABRIC_NOT_READY (919)

Text is copied from the cuda.bindings.driver.CUresult member docstrings
(cuda-bindings==13.4.2), cleaned the same way
DocstringBackedExplanations.get() would clean them.

Bug 6843633

Author: lijinf2

cuda_core/cuda/core/_utils/driver_cu_result_explanations_frozen.py

@@ -1,7 +1,7 @@
 # SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
-# CUDA Toolkit v13.3
+# CUDA Toolkit v13.4
 _FALLBACK_EXPLANATIONS = {
     0: (
         "The API call returned with no errors. In the case of query calls, this"
@@ -54,6 +54,10 @@
         " time. Devices are often unavailable due to use of"
         " ::CU_COMPUTEMODE_EXCLUSIVE_PROCESS or ::CU_COMPUTEMODE_PROHIBITED."
     ),
+    47: (
+        "The API call failed because of a hardware resource required to bind memory to a multicast object is"
+        " unavailable."
+    ),
     100: ("This indicates that no CUDA-capable devices were detected by the installed CUDA driver."),
     101: (
         "This indicates that the device ordinal supplied by the user does not"
@@ -119,6 +123,7 @@
         " work will return the same error. To continue using CUDA, the process must"
         " be terminated and relaunched."
     ),
+    227: "This indicates that the Loader version is insufficient for fatbin",
     300: (
         "This indicates that the device kernel source is invalid. This includes"
         " compilation/linker errors encountered in device code or user error."
@@ -347,5 +352,11 @@
         " with the stream has been destroyed, limiting the stream's operational capabilities."
     ),
     918: "This error indicates that a graph recapture failed and had to be terminated.",
+    919: (
+        "The GPU fabric is not ready within the bounded wait while the fabric manager probe is still in progress (or"
+        " not converging in time). Applications may retry after a delay; for the initialization wait budget, see"
+        " environment variables such as CUDA_FABRIC_INIT_TIMEOUT_MS. The CUDA Runtime uses the same value as"
+        " cudaErrorFabricNotReady."
+    ),
     999: "This indicates that an unknown internal error has occurred.",
 }