Deployment of issues and environment setup

[jdsr1145]

deployment of issue: #3731 #3721 , correction to setup files
fix: pyporject.toml typo in torch version, undistort_view now passes metadata to new camera instance, paralleldatamanager now uses iter_eval_raybundles instead of iter_train_bundles, add a version specification in dockerfile to pass github tests

[Copilot]

Pull request overview

This PR addresses two GitHub issues (#3731 and #3721) by fixing several bugs and correcting version specifications in configuration files. The changes ensure proper metadata propagation in camera transformations, correct dataloader usage in evaluation mode, and update package version constraints.

Changes:

• Fixed torch version specification in pyproject.toml (corrected typo from non-existent version 2.7.1 to >=2.1.2)
• Fixed bug in parallel_datamanager.py where next_eval was incorrectly using training raybundles instead of evaluation raybundles
• Fixed bug in dataloaders.py to preserve camera metadata when creating undistorted camera instances
• Added pip version specification in Dockerfile for CI/CD testing

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File	Description
pyproject.toml	Corrected torch version from invalid "2.7.1" to valid ">=2.1.2"
nerfstudio/data/utils/dataloaders.py	Added metadata parameter to preserve camera metadata in undistort_view function
nerfstudio/data/datamanagers/parallel_datamanager.py	Fixed next_eval to use iter_eval_raybundles instead of iter_train_raybundles
Dockerfile	Added explicit pip version specification (25.2) for GitHub CI/CD tests

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[akristoffersen]

Hi @jdsr1145 , thanks for the contribution. Do we need the torch version to be 2.7.* for anything? Not sure why originally it was set so restrictive.