Unify all external API error message mechanism and enhance third-party API error msg

[zhwesky2010]

PR types

New features

PR changes

Others

Describe

1. 统一所有外部API报错消息机制，使用external_error.proto进行统一管理；

2. 新增强CURAND、CUDNN、CUBLAS、CUSOLVER、NCCL五种Nvidia类型API的报错信息，打印出详细的Hint提示内容，竞品只给出报错码；

---

1). CURAND API:

----------------------
Error Message Summary:
----------------------
OSError: (External)  CURAND error(204). 
  [Hint: 'CURAND_STATUS_ARCH_MISMATCH'.  Architecture mismatch, GPU does not support requested feature. ] (at /workspace/Paddle5/paddle/fluid/platform/gpu_info.cc:99)

2). CUDNN API:

----------------------
Error Message Summary:
----------------------
OSError: (External)  CUDNN error(3), CUDNN_STATUS_BAD_PARAM. 
  [Hint: 'CUDNN_STATUS_BAD_PARAM'. An incorrect value or parameter was passed to the function. To correct, ensure that all the parameters being passed have valid values. ] (at /workspace/Paddle5/paddle/fluid/platform/gpu_info.cc:99)

3). CUBLAS API:

----------------------
Error Message Summary:
----------------------
OSError: (External)  CUBLAS error(13). 
  [Hint: 'CUBLAS_STATUS_EXECUTION_FAILED'. The GPU program failed to execute. This is often caused by a launch failure of the kernel on the GPU, which can be caused by multiple reasons.  To correct: check that the hardware, an appropriate version of the driver, and the cuBLAS library are correctly installed. ] (at/workspace/Paddle5/paddle/fluid/platform/gpu_info.cc:99)

4). CUSOLVER API:

----------------------
Error Message Summary:
----------------------
OSError: (External)  CUSOLVER error(7). 
  [Hint: 'CUSOLVER_STATUS_INTERNAL_ERROR'. An internal cuSolver operation failed. This error is usually caused by a cudaMemcpyAsync() failure.To correct: check that the hardware, an appropriate version of the driver, and the cuSolver library are correctly installed. Also, check that the memory passed as a parameter to the routine is not being deallocated prior to the routine’s completion. ] (at /workspace/Paddle5/paddle/fluid/platform/gpu_info.cc:99)

5). NCCL API:

----------------------
Error Message Summary:
----------------------
OSError: (External)  NCCL error(3), internal error. 
  [Hint: 'ncclInternalError'. An internal check failed. This is either a bug in NCCL or due to memory corruption. ] (at /workspace/Paddle5/paddle/fluid/platform/gpu_info.cc:99)

6). CUDA API:

共包括116种错误类型与报错信息，不进行改动，在1.8已经进行支持；

---

缺省信息:

上述所有API如果无法查找到爬虫中对应的错误码，将会默认打印：

----------------------
Error Message Summary:
----------------------
OSError: (External)  CUDNN error(100). 
  [Hint: Please search for the error code(100) on website(https://docs.nvidia.com/deeplearning/cudnn/api/index.html#cudnnStatus_t) to get Nvidia's official solution and advice about CUDNN Error. ] (at /workspace/Paddle5/paddle/fluid/platform/gpu_info.cc:99)


注：CUDNN仅为示例，实际会根据External API的类型来返回对应网址；

[paddle-bot-old]

Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

[chenwhql]

2020->2021?

[chenwhql]

why not need verison now?

[zhwesky2010]

现在分了6种API，每种API取了一个网址，如果再分小版本就太多了。因为低版本API都是包含在高版本里，现在用的CUDA11.2 是比较全的

[chenwhql]

CudaStatusType命名是不是不太严谨，用NvidiaLib或者别的？

[zhwesky2010]

这个地方是写错了吗

写错了，改过来了。

[zhwesky2010]

用了ExternalApiType，看是用NvidiaApiType还是？

[chenwhql]

这个地方是写错了吗

[chenwhql]

LGTM