Matt McKay

@longye-tian can you take this issue on?

@HumphreyYang I know there was discussion in https://github.com/QuantEcon/lecture-python-intro/pull/270 around fixing this exercise. Would you mind to review this thread and create a PR?

@shlff is this still `in-work`?

thanks @shlff it sounds like our cloud machine needs more `hdd` so I will look into this.

Current our docker container is working well. So this is a low priority switch. https://github.com/QuantEcon/lecture-python.docker

@shlff so it seems the `filesystem` size is to correct. I did a bunch of testing and there is plenty of file system memory. It looks like this is closer...

- [x] remove google collab testing and open a new PR to reinstate (with new testing). It looks like the docker container for google collab isn't properly configured for nvidia...

- [x] re-enable build cache - [x] remove `docker` dependency and test local builds using `anaconda` (simpler)

- [x] remove `cake_eating_numercial` if timing doesn't work in the preview (as per comment). This is captured in https://github.com/QuantEcon/lecture-jax/issues/185

@jstac this is looking good for your review and merge.