Make WebHDFS ls() method compatible with superclass by supporting **kwargs

[mazurov]

The current WebHDFS.ls() implementation does not accept additional parameters, which breaks compatibility with newer versions of JupyterFS (e.g., jupyterfs==1.1.0).

Since we abstract filesystem interfaces, all common methods (such as ls, info, etc.) should accept **kwargs to ensure forward compatibility with upstream fsspec and JupyterFS expectations.

For example, JupyterFS 1.1.0 calls fs.ls(..., refresh=True) internally, which raises:

TypeError: WebHDFS.ls() got an unexpected keyword argument 'refresh'

Reference:
https://github.com/jpmorganchase/jupyter-fs/blob/6fac72f9a047871de48b757c5634317f88b5ac77/jupyterfs/manager/fsspec.py#L188

Proposed fix:

• Match the signature of the base class, which accepts **kwargs.
• Update WebHDFS.ls() signature to accept **kwargs as it implemented for other filesystems (e.g. local)

[martindurant]

@RRosio : do you think it's reasonable that jupyter-fsspec should be calling ls(.., refresh=True)? Rather than changing the signatures for filesystems as suggested here, it would be easier to use use_listings_cache=False to the constructor, or call .invalidate_cache() before listing.

However, should we not leave it to the user configuration whether they want directory caching or not?

[martindurant]

@mazurov , notwithstanding my comment above, your change is still worthwhile, because the superclass signature is

def ls(self, path, detail=True, **kwargs)

[mazurov]

@martindurant, I have to change the original objective from support jupyterfs to follow the signature of the base class