[caffe2] fix ambiguous call to 'fmaxType' THCHalfAutoNumerics.cuh (#33569)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/33569
Clang reported a few places where a call to `fmaxType` is ambiguous. In all cases one of the arguments is `double` and another is `float`. Fix the error by creating a proper value 0 and remove the unneeded `ZERO_MACRO` code.
Test Plan:
```lang=bash
buck build mode/opt -c fbcode.cuda_use_clang=true //fblearner/flow/projects/dper:workflow
buck build mode/opt //fblearner/flow/projects/dper:workflow
```
Execute tests on devgpu:
```
buck test mode/dev-nosan -j 8 //caffe2/caffe2/python/operator_test/... //caffe2/test:cuda
```
Reviewed By: ngimel
Differential Revision: D20006926
fbshipit-source-id: ca6cfacd57459b1c48eb5080b822d9509b03544d
