[GraphBolt] ndata/edata adding for heterograph when preprocessing

## 🐛 Bug

## To Reproduce

when adding feature data into nodes/edges for heterograph, `ntype` or `etype` is not used.
https://github.com/dmlc/dgl/blob/d82cfdc05c9be7dc15bd77e7ceec613f48fcf888/python/dgl/graphbolt/impl/ondisk_dataset.py#L143
https://github.com/dmlc/dgl/blob/d82cfdc05c9be7dc15bd77e7ceec613f48fcf888/python/dgl/graphbolt/impl/ondisk_dataset.py#L150

## Expected behavior

## Environment

- DGL Version (e.g., 1.0):
- Backend Library & Version (e.g., PyTorch 0.4.1, MXNet/Gluon 1.3):
- OS (e.g., Linux):
- How you installed DGL (`conda`, `pip`, source):
- Build command you used (if compiling from source):
- Python version:
- CUDA/cuDNN version (if applicable):
- GPU models and configuration (e.g. V100):
- Any other relevant information:

## Additional context

Start in python/dgl/graphbolt/impl/ondisk_dataset.py at the referenced lines 143 and 150, where feature data is added for nodes and edges. Trace how heterograph node and edge types are represented during preprocessing, then reproduce the case with multiple node or edge types. Done means ntype and etype are respected when adding the corresponding feature data.

Written by the indexing model from the issue text.