Generation ID ignored when reading from gcs

Hi, I am having an issue reading specific generation/versions of files from gcs.

For instance, in a bucket with versioning enabled:

# old
gs://mock-bucket/something.csv#123456789
# new
gs://mock-bucket/something.csv#999999999

when trying to read gs://mock-bucket/something.csv#123456789, it will read gs://mock-bucket/something.csv#999999999 without any warning or error message.

I am happy to make a contribution! let me know!

To reproduce:

• Create a gcs bucket with version enabled.
• Upload two file with different content to the bucket with the same key.
• This will generate 2 objects with same key but they will have different generation id
• use ffspec to read the object with older generation id
  • ie:

  from ffspec import open as f_open
  
  with f_open.read('gs://mock-bucket/something.csv#123456789', 'r') as f:
      ....   
  
  

• ffspec will only read ignore the generation id and only read the latest files.

Start by tracing the fsspec.open entry point for gs:// URLs and how the generation fragment is handled. Reproduce the case with a versioned Google Cloud Storage bucket; done means reading an older generation returns that content while requests without a generation still read the latest object.