Fix pretraining_transforms for ImageEmbedder

[krshrimali]

What does this PR do?

Attempts to fix #1185. Before this, pretraining_transforms was never called.

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests? [not needed for typos/docs]
• Did you verify new and existing tests pass locally with your changes?
• If you made a notable change (that affects users), did you update the CHANGELOG?

PR review

• Is this pull request ready for review? (if not, please submit in draft mode)

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

[codecov]

Codecov Report

Merging #1196 (bb27c2a) into master (36377b7) will increase coverage by 0.12%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #1196      +/-   ##
==========================================
+ Coverage   90.99%   91.11%   +0.12%     
==========================================
  Files         284      284              
  Lines       12740    12755      +15     
==========================================
+ Hits        11593    11622      +29     
+ Misses       1147     1133      -14     

Flag	Coverage Δ
unittests	91.11% <100.00%> (+0.12%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
flash/core/data/io/input_transform.py	72.72% <100.00%> (+0.13%)	⬆️
flash/image/embedding/heads/vissl_heads.py	100.00% <100.00%> (ø)
flash/image/embedding/losses/vissl_losses.py	100.00% <100.00%> (+8.33%)	⬆️
flash/image/embedding/model.py	91.22% <100.00%> (+2.76%)	⬆️
flash/image/embedding/vissl/adapter.py	89.01% <100.00%> (+0.77%)	⬆️
flash/image/embedding/backbones/vissl_backbones.py	74.07% <0.00%> (-7.41%)	⬇️
flash/core/adapter.py	98.38% <0.00%> (+1.61%)	⬆️
...ash/image/embedding/strategies/vissl_strategies.py	100.00% <0.00%> (+8.57%)	⬆️
... and 2 more

---

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 36377b7...bb27c2a. Read the comment docs.

[krshrimali]

Update: looks like it's still not called. WIP :(

[ethanwharris]

Awesome!!! LGTM, just a few small comments

[ethanwharris]

Nice!

[krshrimali]

@ethanwharris - How to get rid of codecov failures? Can you please help with that?

Also, there are these tests failing: https://dev.azure.com/PytorchLightning/lightning%20Flash/_build/results?buildId=59765&view=logs&j=fb683405-d979-52da-6de9-2541dff429a6

# Failed while doing: nltk.download("punkt", quiet=True, force=False)

With an error:

Resource �[93mpunkt�[0m not found.
2022-03-03T04:45:07.0342026Z   Please use the NLTK Downloader to obtain the resource:

And the suggested fix from the CI:

# suggestion from the error:
import nltk
nltk.download('punkt')

Should we try downloading punkt with the CI config for azure?

[krshrimali]

Thank you @ethanwharris for fixing the failures! All the tests pass now 🚀, regarding the DeepSource failures, I think we can safely ignore them for now? (is there a way to auto-merge with these failures?)

I have a minor suggestion though: https://deepsource.io/gh/PyTorchLightning/lightning-flash/run/cc57fc8b-c459-4137-8c24-69f7ddc9b119/python/PYL-W0221 - I think deepsource is right here, and we can make loss_fn, backbone, head, hooks as kwargs (it won't be a BC breaking change) in a separate PR, to make sure that we override correctly. What do you think?