Improve error handling for URL requests in dataset fetch methods

[ascillitoe]

The requests.get() calls in the fetch methods such as fetch_nab and fetch_cifar10c currently don't have any error handling, potentially resulting in unclear errors when the requested url is down. For example see:

https://github.com/SeldonIO/alibi-detect/blob/436e99efb88c922dc04cec97b0b52d88bf439d65/alibi_detect/datasets.py#L369-L371

The url requests could be wrapped in try/except blocks which log an exception message and raise an error, in a similar way to the alibi datasets methods:

https://github.com/SeldonIO/alibi/blob/3a589c4de86a89fbdb956427d0d3020de05bdfda/alibi/datasets.py#L105-L110

Logic is also needed in the tests to skip tests when the URL is down, for example:

https://github.com/SeldonIO/alibi/blob/3a589c4de86a89fbdb956427d0d3020de05bdfda/alibi/tests/test_datasets.py#L13-L17

[ascillitoe]

Coming back to this. The only issue with skipping the test when a URL is down is that CI won't pick up cases where URL's are permanently dead or changed. Ideally we need to skip temporarily dead URL's during nightly and PR tests, but still check these less frequently i.e. during CI for releases.

[jklaise]

Fixed by #350.

[ascillitoe]

Reopening this since we missed a utility function in alibi_detect.datasets; the corruption_types_cifar10c function causes a CI failure if https://storage.googleapis.com/seldon-datasets is down (see https://github.com/SeldonIO/alibi-detect/runs/6662544833). We should add the try-except logic added in #350 to any tests using this function.