Make pft.imsave replacebly by skimage.io imsave

[haesleinhuepf]

A common usecase would be that someone writes a notebook on their laptop with

from skimage.io import imsave

Then, they want to run it on the cluster, where the code would be

pft = bt.ProjectFileTransfer(source_mount, target_project)
imsave = pft.imsave

It would be cool if this was simpler...

[haesleinhuepf]

Actually, that's a cool idea, because also this would work:

from pandas import from_csv

and alternatively:

pft = bt.ProjectFileTransfer(source_mount, target_project)
from_csv = pft.pandas_from_csv

[haesleinhuepf]

Alternatively, if this would work, we could switch such an environment:

bt.ProjectFileTransfer("C:/temp", "C:/Robert/Data")

... I think that's the most convenient way

[thawn]

the function signatures of the various save and read functions are now the same as for the original functions.
i.e. I removed the target parameter from the save functions (and replaced it by an attribute of projectFileTransfer).
This means that now something like
imsave = pft.imsave should work (need to do more testing, though)

In principle, this should even work locally, because pft always looks if it finds the file locally before trying to download it from the fileserver. Also, pft saves the file locally, if it can write to the parent directory (except if the parent directory is the current working directory)

however, there are a few open tasks:

• handle automatic conversion of windows to linux paths (use pathlib for that)
• how to deal with saving files to the current working directory? We almost always have write access there, but what about if the user wants to save to the fileserver instead? (therefore, we currently save to the fileserver/project space instead of the current working directory) -> configure when initializing the pft object
• how to deal with pandas.to_x methods? If we really wanted this to work as a plug-in replacement, we would have to implement our own pandas dataframe, that inherits from the original and overrides those methods. - a bit daunting (because pandas has so many of these methods and we would have to override them all) -> leave this as is for now, until it is needed