[AL-5278] [AL-5279] [AL-5260] Add catalog slice, data row, dataset support

[mnoszczak]

AL-5278
AL-5279
AL-5280

Edit: Added test cases for all the methods, no need to run the below

Need to prepare staging acc to add tests, right now can be tested by checking out to the branch and:

pip install -e .  

from labelbox import Client
API_KEY = API_KEY
ENDPOINT = STAGE_URL
SLICE_ID = 

client = Client(api_key=API_KEY, endpoint=ENDPOINT)
slice = client.get_catalog_slice(SLICE_ID)
task = slice.export_v2(params={"performance_details": False, "label_details": True})

task.wait_till_done()
task.result

[mnoszczak]

We have schema checker that expects the query to be in the first line

[dubininsergey]

The code changes look good to me, works fine locally 👍
Would be great to refactor it, but no-blocker

[dubininsergey]

should we cleanup this prints and leverage logger.* where needed? Eg. logger.debug or logger.info

[dubininsergey]

A minor suggestion: these 3 export_v2 methods could be moved to a separate dedicated service like ExportService. This service could have 3 methods: export_dataset, export_data_row and export_slice. And then these methods could be called in the corresponding classes (Dataset, DataRow and Slice).

This way you would be able to reuse the code parts that are currently duplicated in Dataset.export_v2, DataRow.export_v2 and Slice.export_v2: fetching the task object or building and calling exportDataRowsInCatalog(though it's duplicated only in DataRows and Dataset classes).

Besides reducing code duplication it would increase coherence of the code in ExportService. Also Datasetclass will get to 723 lines long size with this change, putting the most of logic into ExportService would avoid inflating the modules' sizes

[mnoszczak]

That was my thought as well — will refactor this as part of SDK v4 initiative that aims to use REST endpoints for SDK instead (we'll need backend changes to facilitate this).

[tytalus]

lgtm