[bug fix] fix fp16 dtype checking for argmax op

[zhangting2020]

PR types

Bug fixes

PR changes

APIs

Describe

• good first issue for fp16 #50832 case1

问题描述：在静态图模式下，输入为FP16类型时，argmax会报TypeError。

import paddle
import numpy as np

paddle.enable_static()

x_np = np.random.random((10, 16)).astype('float16')
x = paddle.static.data(shape=[10, 16], name='x', dtype='float16')
out = paddle.argmax(x)

exe = paddle.static.Executor()
exe.run(paddle.static.default_startup_program())
out = exe.run(feed={'x': x_np},
            fetch_list=[out])

报错：

Traceback (most recent call last):
  File "test_argmax.py", line 8, in <module>
    out = paddle.argmax(x)
  File "/usr/local/lib/python3.7/dist-packages/paddle/tensor/search.py", line 201, in argmax
    'paddle.argmax',
  File "/usr/local/lib/python3.7/dist-packages/paddle/fluid/data_feeder.py", line 115, in check_variable_and_dtype
    check_dtype(input.dtype, input_name, expected_dtype, op_name, extra_message)
  File "/usr/local/lib/python3.7/dist-packages/paddle/fluid/data_feeder.py", line 191, in check_dtype
    extra_message,
TypeError: The data type of 'x' in paddle.argmax must be ['float32', 'float64', 'int16', 'int32', 'int64', 'uint8'], but received float16.

修复方案：

• 在argmax API静态图模式下的类型检查中增加fp16支持，并且修改API doc
• 修复对应的API中文文档：support float16 for argmax docs#5654

[paddle-bot]

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

[AnnaTrainingG]

LGTM