feat(loss)/add different operator types for cross_entropy

[yingtongxiong]

Motivation

1. support different operator types for cross_entropy
   
2. add op_type config
   loss = dict( label_smoothing=0, op_type="flash_vocab_parallel", )

reference this https://github.com/pjlab-sys4nlp/InternTrain/pull/148

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here and update the documentation.

Checklist

Before PR:

• Pre-commit or other linting tools are used to fix the potential lint issues.

[sunpengsdu]

@yingtongxiong merge下develop分支过CI吧

[yingtongxiong]

@yingtongxiong merge下develop分支过CI吧

好的

[yingtongxiong]

@yingtongxiong merge下develop分支过CI吧

@sunpengsdu 已经好了