Python: Use Version Ranges for Various Dependencies

[srilman]

As discussed in #6620, this PR uses version ranges for some dependencies. This will unfix the versions for certain dependencies and allow users to use older versions when install PyIceberg to use as a library. The dependencies changed are:

• PyArrow from 11.0.0 to 8.0.0-11.0.0. It is an optional dependency used either for FileIO, conversion from Iceberg to Arrow (schema, expression, table, etc.), or a dep for Pandas or DuckDB. Made sure current tests pass with every major version.
• Pandas from 1.5.3 to 1.4.4-1.5.3. It is only used to read Iceberg tables out to Pandas, which is current done through PyArrow anyways. Tested 1.4.4. PyIceberg used 1.5.2 until recently
• DuckDB from 0.6.1 to 0.6.0-0.6.1. Similar to Pandas, and PyIceberg used 0.6.0 until recently
• FsSpec, S3Fs, and ADLSFs from 2023.1.0 to 2022.8.2-2023.1.0. Used for FileIO, and I believe S3Fs and ADLSFs already depend on the same version of FsSpec. Tested with all major versions with success, and PyIceberg has used 2022.8.2, 2022.10.0, and 2022.11.2 in the past
• Requests from 2.28.2 to 2.28.1-2.28.2 since this is only a patch release and PyIceberg used 2.28.1 until recently.
• Rich from 13.3.1 to 13.0.0-13.3.1 since this is minor release changes with no significant differences
• PyYAML from 6.0.0 to 5.4.0-6.0.0 since 5.4 is still used commonly in other tools, and the only change in the major change in the major release was removing Python 2.7 support. Tested with 5.4.0

[srilman]

Ready for review. @Fokko or someone else?

[Fokko]

Thanks for pinging me @srilman and sorry for the late review. This looks great, thanks for going over this! 👏🏻

[Fokko]

Would be cool if we can get some kind of TOML linter.