Dataset cache design

[wang3702]

For the code here https://github.com/wfondrie/depthcharge/blob/486221c6fd8cb7cda68ac7ec0996e3b3dfb6c88a/depthcharge/data/spectrum_datasets.py#L108C9-L117C14.

I don't understand why don't we just simply use the _path's cached lance dataset if it is specified by the parameters. Instead, we overwrite the existing cached dataset. I think it is better to judge if this path is specified by the user or not. If yes, it is better to directly load it to save time.