[Tokenizer] Unify tokenizer _pad

[DrownFish19]

PR types

Function optimization

PR changes

APIs

Description

Unify tokenizer _pad function.

• Move the attention_mask([1,seq_len,seql_len]) padding action into tokenizer_base _pad.
• Move attn_mask_startend_row_indices padding action into tokenizer_base _pad.

[FlashMask] Add FlashMask for Qwen2 #9264误差范围验证基于此PR

[paddle-bot]

Thanks for your contribution!

[codecov]

Codecov Report

Attention: Patch coverage is 92.30769% with 2 lines in your changes missing coverage. Please review.

Project coverage is 52.96%. Comparing base (1dc6e18) to head (5ad7381).
Report is 2 commits behind head on develop.

Files with missing lines	Patch %	Lines
paddlenlp/transformers/tokenizer_utils_base.py	87.50%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #9280      +/-   ##
===========================================
+ Coverage    52.94%   52.96%   +0.02%     
===========================================
  Files          657      657              
  Lines       106533   106384     -149     
===========================================
- Hits         56404    56351      -53     
+ Misses       50129    50033      -96     

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

[lugimzzz]

LGTM