Fix: Data and HeteroData feature store get_all_tensor_attrs

[mananshah99]

Resolves a bug in Data and HeteroData which results in get_all_tensor_attrs not filtering for true tensor attributes.

[codecov]

Codecov Report

Merging #4882 (0b4894f) into master (cfda074) will increase coverage by 0.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #4882      +/-   ##
==========================================
+ Coverage   82.76%   82.77%   +0.01%     
==========================================
  Files         329      329              
  Lines       17703    17704       +1     
==========================================
+ Hits        14652    14655       +3     
+ Misses       3051     3049       -2     

Impacted Files	Coverage Δ
torch_geometric/data/data.py	91.43% <100.00%> (ø)
torch_geometric/data/hetero_data.py	94.13% <100.00%> (+0.01%)	⬆️
torch_geometric/data/storage.py	81.64% <0.00%> (+0.63%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update cfda074...0b4894f. Read the comment docs.

[yaoyaowd]

Thanks!

[rusty1s]

@mananshah99 I just had a second look and this will somehow break PyG compatibility once we make the full switch to FeatureStore: Currently, every attribute will just get passed down as a reference to the mini-batch object (without indexing), in case it is neither a node or edge tensor. Let me know what you think!

[rusty1s]

@mananshah99 Shall we create an issue for that?