Test visualisations in Colab

[kwabenantim]

• Build a new updated conda package with the latest visualisation fixes from fix PYBIND11_VTK_TYPECASTER macro and supply typecasters for vtkRende… #5
• Test in Colab
• Fix any issues

[kwabenantim]

Problem: scene.Start()

• Kernel survives in Jupyter but dies in Colab after scene.Start():

• The Colab error is:

INFO	KernelRestarter: restarting kernel (1/5), keep random ports
WARNING	vtkXOpenGLRenderWindow (0x3f627c0): bad X server connection. DISPLAY=
WARNING	ERROR: In /home/conda/feedstock_root/build_artifacts/vtk_1686064901542/work/Rendering/OpenGL2/vtkXOpenGLRenderWindow.cxx, line 300
WARNING	WARNING:root:kernel 4b3cf4d0-ff5b-4d11-9289-58e8acea4a0c restarted

Fix

# From same issue in https://github.com/Kaggle/docker-python/pull/360
!apt-get install -y xvfb
!mamba install -c conda-forge xvfbwrapper
from xvfbwrapper import Xvfb
vdisplay = Xvfb()
vdisplay.start()
render()
vdisplay.stop()

• Moving to an osmesa-based vtk build would be a cleaner future solution although it might require more changes to cmake config etc.

[kwabenantim]

Problem: scene.SetSaveAsAnimation(True)

• Kernel dies in locally compiled, pip-installed chaste in Jupyter notebook after:

scene.SetSaveAsAnimation(True)
scene.Start()

• The Jupyter error is:

[I 18:16:30.536 NotebookApp] KernelRestarter: restarting kernel (1/5), keep random ports
WARNING:root:kernel 94d14e59-9a06-474e-bf03-5315bf2f7990 restarted
[IPKernelApp] WARNING | debugpy_stream undefined, debugging will not be enabled

• The Xvfb solution above doesn't fix this.
• Possibly due to system vtk.
• The conda package runs fine in local Jupyter.

Fix:
Fixed by making JupyterNotebookManager a singleton. See #1 for related memory leak issues.