[INFER][LLM] Add the AutoPredictor for inference

[zeroRains]

PR types

New features

PR changes

Others

Description

整合现有全部Predictor的选择条件，创建AutoPredicotr，可以处理static graph/dynamic graph，inference_mode/not inference_mode，block_attn/not block_attn的情况。

调用方式

predictor = AutoPredictor.create_predictor(predictor_args, config, model_args, tokenizer, model=model)

AutoPredictor的执行流程图如下：

当前LLM最简单的调用流程如下图：

[paddle-bot]

Thanks for your contribution!

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 3 lines in your changes missing coverage. Please review.

Project coverage is 53.08%. Comparing base (4b02477) to head (42b3e6a).
Report is 229 commits behind head on develop.

Files with missing lines	Patch %	Lines
paddlenlp/transformers/auto/modeling.py	0.00%	3 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #9445      +/-   ##
===========================================
+ Coverage    52.93%   53.08%   +0.14%     
===========================================
  Files          688      699      +11     
  Lines       109379   110476    +1097     
===========================================
+ Hits         57899    58641     +742     
- Misses       51480    51835     +355     

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