Consider updating jupyter-bokeh dependency from ~=3.0.5 to support 4.x series

[coderabbitai]

Background

We're working on aligning Jupyter dependencies across OpenDataHub workbench notebook images and noticed that TrustyAI currently constrains jupyter-bokeh to ~=3.0.5 in both requirements.txt and pyproject.toml.

Current Situation

• TrustyAI constraint: jupyter-bokeh~=3.0.5
• Other workbench images: Moving to jupyter-bokeh~=4.0.5
• Impact: TrustyAI workbench is currently the only image that cannot upgrade

Request

We'd appreciate guidance on the feasibility of updating the jupyter-bokeh dependency to support the 4.x series (specifically ~=4.0.5). This would help maintain dependency consistency across our workbench ecosystem.

Important Notes

⚠️ This is NOT urgent - we're not experiencing any specific problems with the current 3.x constraint. This is purely a dependency alignment effort to keep all workbench images in sync.

@ruivieira - Could you provide insight on how difficult this upgrade might be from TrustyAI's perspective? Any known compatibility concerns or breaking changes we should be aware of?

Technical Context

The jupyter-bokeh 4.x series includes some breaking changes compared to 3.x:

• Bokeh API changes (ColorMap → PaletteSelect widget rename)
• Migration from ES2017 to ES2020 JavaScript standards
• Some widget behavior changes

References

• Related PR: RHOAIENG-21842: bump JupyterLab-related dependencies in all Python 3.11 and 3.12 Pipfiles opendatahub-io/notebooks#1218
• Discussion: RHOAIENG-21842: bump JupyterLab-related dependencies in all Python 3.11 and 3.12 Pipfiles opendatahub-io/notebooks#1218 (comment)
• Requested by: @jiridanek

Thank you for considering this request! 🙏

[ruivieira]

@jiridanek This was fixed with #224, but we'll probably batch some more dependency changes for the upcoming 0.6.3 release.