Remove upper bound for fsspec

[fellhorn]

Describe the bug

As also raised by @cyyever in #7296 and @NeilGirdhar in d546883#commitcomment-140952162 , datasets has a problematic version constraint on fsspec.

In our case this causes (unnecessary?) troubles due to a race condition bug in that version of the corresponding gcsfs plugin, that causes deadlocks: fsspec/gcsfs#643

We just use a version override to ignore the constraint from datasets, but imho the version constraint could just be removed in the first place?

The last few PRs bumping the upper bound were basically uneventful:

• bump fsspec #7219
• Support fsspec 2024.5.0 #6921
• chore(deps): bump fsspec #6747

Steps to reproduce the bug

Expected behavior

Installing fsspec>=2024.10.0 along datasets should be possible without overwriting constraints.

Environment info

All recent datasets versions

[lhoestq]

Unfortunately fsspec versioning allows breaking changes across version and there is no way we can keep it without constrains at the moment. It already broke datasets once in the past. Maybe one day once fsspec decides on a stable and future proof API but I don't think this will happen anytime soon

edit: bumped to 2024.10.0 in #7352