Fix Image Feature Extraction Observation by Removing Double Normalization and Using Only GPU Operations

[glvov-bdai]

Description

Previously, image_features requested normalized images prior to normalizing as part of the preprocessing. As pointed out to me by Jinghuan Shang (Theia Vision Transformer first author), this lead to images being normalized more than once which is ineffective and incorrect. Furthermore, Jinghuan pointed out where Transformers needlessly converted from GPU to CPU for normalization, where these operations could be converted to be GPU only.

ALSO: Fixes device and setup.py

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)

Checklist

• I have run the pre-commit checks with ./isaaclab.sh --format
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• I have updated the changelog and the corresponding version in the extension's config/extension.toml file
• I have added my name to the CONTRIBUTORS.md or my name already exists there

[jsmith-bdai]

LGTM thanks!

[Mayankm96]

Shall we close this MR in favor of #1340 ?

[glvov-bdai]

Yup works for me. Closed in favor of #1340