Added creation and preview of local image datasets with in place caption editing and searching abilities

[Sourenm]

Users can create either text or image datasets,
They can drag and drop or choose any number of files to upload.,
One (or the first one) of these files could be json, jsonl, or csv that is the metadata for the images.,
If no such file is provided the app will create an empty metadata file itself,
User can view the images and the captions of the created dataset using the preview button,
The preview allows for searching based on the caption, and in place editing of the captions which will get written back to the metadata file once the save changes button is pressed

[deep1401]

This modifies the DataTemplateWithPreview tab which would break the training task (Data Template) window. Maybe we can add these things in some newer file instead?