Data pipeline refactoring

[hadipash]

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

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

Motivation

Refactored data pipeline to match best MindData practices, including:

1. Use GeneratorDataset for data loading only.
2. Use dataset.map operation to apply data transformations and augmentations.
3. Reduce number of Python transformations by grouping them into a single operation.
4. Group MindSpore operations as well.
5. Move to MindSpore operations where it is possible (Decode, Normalize, HWC2CHW).
6. Integrate MindRecord support.

[hadipash]

Rebased onto the main branch to resolve conflicts.