update dev before release

[aliutkus]

• I think that having the functional interface lie only in some predict.py file is not so much intuitive => I imported separate in the __init__ of the package along with the models constructions. This provide a functional openunmix.separate function. I believe it's way simpler and better this way
• I really think this functional interface should take both tensors or paths as inputs. the separate function was limited to filepaths. I think that when you call separation programatically, it's actually quite common that you want to process tensors directly. I modified the separate function so that you can either have input as a str (in which case it's understood as a filepath) or as a torch.Tensor (in which case you should also provide the otherwise ignored rate parameter)
• I don't think there was any need for having some predict.inference_args. I duplicated the corresponding code in cli.py and eval.py I understand it's arguably less elegant, but I'm actually not even sure we want the same defaults for cli and eval. It's just 4 parameters anyways, and I believe it makes the code more readable.

[codecov[bot]]

Codecov Report

Merging #65 into dev will increase coverage by 1.82%. The diff coverage is 21.42%.

@@            Coverage Diff             @@
##              dev      #65      +/-   ##
==========================================
+ Coverage   63.87%   65.69%   +1.82%     
==========================================
  Files          19       18       -1     
  Lines        1204     1169      -35     
==========================================
- Hits          769      768       -1     
+ Misses        435      401      -34     

Impacted Files	Coverage Δ
openunmix/evaluate.py	39.34% <0.00%> (-3.76%)	:arrow_down:
setup.py	0.00% <ø> (ø)
openunmix/predict.py	19.04% <25.00%> (-2.70%)	:arrow_down:
openunmix/__init__.py	40.90% <100.00%> (+0.90%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update a89052b...d8ba6e3. Read the comment docs.