[BUG] avalanche-lib import fails

[LaurenzBeck]

🐛 Bug Report

🔥 Current Behavior

when importing streamgen.streams, the imports of from avalanche.benchmarks.utils import as_classification_dataset cause an import error due to a mistake in the version constraints. I checked the responsible torchcv packe and it does not adhere to the necesary standards to solve this issue with a reasonable amount of effort.

Because I do not want to wait for the fix and since the avalanche helper mehtods only provide application-specific starting points, I decided to exclude them temporarily.

✨ Expected Behavior

I want from avalanche.benchmarks.utils import as_classification_dataset to work on a fresh install 😅

💻 Reproducing the Bug

In the environment, simply call from avalanche.benchmarks.utils import as_classification_dataset

[LaurenzBeck]

Since I only need very small parts of avalanche and continuum, I am thinking of:

• copying the corresponding source files to streamgen to avoid the heavy dependencies
• create a separate application environment for the examples and remove the pip example extra