310p export bug fix

[kk928290341]

Thank you for your contribution to the MindOCR repo.
Before submitting this PR, please make sure:

• You have read the Contributing Guidelines on pull requests
• Your code builds clean without any errors or warnings
• You are using approved terminology
• You have added unit tests

Motivation

310p机器ocr检测和识别模型 全量export，发现可修复的bug

1. crnn_resnet34_ch convert shape匹配不上
2. 文件存在时remove报错
3. Abinet, svtr_tiny export失败解决：
   (1) 有些ms版本在init里不能有tensor操作，一般是要先用numpy算好，再转成Tensor
   (2)矩阵乘维度对不上，没有自动广播

(Write your motivation for proposed changes here.)

Test Plan

(How should this PR be tested? Do you require special setup to run the test or repro the fixed bug?)

Related Issues and PRs

(Is this PR part of a group of changes? Link the other relevant PRs and Issues here. Use https://help.github.com/en/articles/closing-issues-using-keywords for help on GitHub syntax)

[horcham]

涉及算子修改，需复验精度。验证精度用以下命令：

rec类模型(例如svtr_tiny)：
python tools/benchmarking/multi_dataset_eval.py --config configs/rec/svtr/svtr_tiny.yaml

det类模型（例如dbnet_r50)：
python tools/eval.py --config configs/det/dbnet/db_r50_icdar15.yaml

[kk928290341]

你发的邮件我已收到，谢谢

[kk928290341]

复验精度
abinet 平均acc=91.35%
svtr-tiny 平均acc=90.25%
达标

[Mark-ZhouWX]

右上角修改一下Reviewers

[Mark-ZhouWX]

approve

[kk928290341]

你发的邮件我已收到，谢谢