[Feature] Refactor visual grounding codes, support format-only inference and evaluation scripts

[Tai-Wang]

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

Please describe the motivation of this PR and the goal you want to achieve through this PR.

• Support inference and evaluation related codes for the multi-view 3D visual grounding challenge.

Modification

Please briefly describe what modification is made in this PR.

• Refactor the visual grounding dataset class to make it more compatible with EmbodiedScanDataset.
• Remove the dependence on positive_tokens/maps during format-only inference.
• Support format_only argument in the GroundingMetric.
• Add submission script under tools to support converting the saved predictions to the submission format.
• Add evaluation script under tools to support evaluating the prediction file under the submission format.

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
3. If the modification has potential influence on downstream projects, this PR should be tested with downstream projects.
4. The documentation has been modified accordingly, like docstring or example tutorials.