[LLM INFER] top_p_sampling_reject support top_p=0 and custom seed

[gzy19990617]

PR types

Bug fixes

PR changes

Others

Description

1.支持传入seed，支持不同batch的top_p val不同。

2.该top_p_sampling_reject算子在TopP的值为0的时候，无出现无法采样成功的问题，返回值不正确。解决办法，在kernel内部进行修改，TopP为0时，采用TopK，同时进行了一版优化，将一些同步与等待消除掉。

3.fix一些之前未测试出来的小bug。

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 2 lines in your changes missing coverage. Please review.

Project coverage is 53.02%. Comparing base (8b2cc08) to head (c1c7bc7).
Report is 267 commits behind head on develop.

Files with missing lines	Patch %	Lines
...enlp/experimental/transformers/generation_utils.py	0.00%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #9202      +/-   ##
===========================================
- Coverage    53.16%   53.02%   -0.15%     
===========================================
  Files          657      657              
  Lines       106311   106311              
===========================================
- Hits         56524    56367     -157     
- Misses       49787    49944     +157     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.