xiang song(charlie.song)

*Issue #, if available:* *Description of changes:* By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

The V0.3 release will focus on optimizing the document of GraphStorm. It will also includes other improvements and features. - [Improvement] GraphStorm Documentation - [x] Update document of GSProcessing -...

Based on our DGL-KE project, adversarial negative loss is useful to improve link prediction model performance. https://github.com/awslabs/dgl-ke/blob/43875a57a721d62396d05235226b13e6c2db1a2a/python/dglke/models/pytorch/loss.py#L85-L90

*Issue #, if available:* *Description of changes:* Add support for edge feat reconstruct in multitask learning. We treat edge feature reconstruction as an regression task. By submitting this pull request,...

We need to add multitask training/inference data processing support in GSProcessing. ```[tasklist] ### Tasks - [ ] https://github.com/awslabs/graphstorm/issues/992 - [ ] https://github.com/awslabs/graphstorm/issues/993 ```

We should support loading existing feature transform rules for floating point features in GSProcessing.