Update azure-core dependency version constraint

[benjaminkaplan]

There are 2 version naming schemes that are used by Azure (real version and by date). This causes many issues with conda environments.

I outlined the issue here and here. Simply removing this constraint would fix the issue.

There is precedent for this. Look at azure-storage-blob. It does not have a maximum version, which allows it to be installed from either azure-storage=2025.09.01 available here or from the more specific azure-storage-blob=12.27.1 available here.

Making this change would do a similar things for azure-core=2025.09.01 (here) and azure-core=1.36.0 (here)

[benjaminkaplan]

Just wanted to add, that you can see here that azure-core=1.36.0 is included inside of azure-core=2025.09.01. So there should be no compatibility issues.

[kyleknap]

Hi @benjaminkaplan. Thanks for submitting the PR. Let's continue the discussion on what we want to do in the other GitHub issue you opened: #522.

Also, if we want to make a change in adlfs, I suspect that we will need to update the pyproject.toml or where the adlfs conda recipe is defined as the files in the requirements directory is really only for use in CI and not in packaging adlfs. But before making any more updates to this PR, I want to make there's alignment on understanding of the issue/solution.

[benjaminkaplan]

@kyleknap
Understood. I added an explanation and an example here