ci: proper setting oldest/latest for integrations & PL>=1.9

[Borda]

What does this PR do?

Integrations on GPU were strangely set always oldest req.

Dropped compatibility with:

• 1.6 - no special reason, but the next two versions wont work so skip also this old one
• 1.7 - Pl was looking for removed protected function, ERROR integrations/test_lightning.py - ImportError: cannot import name '_compare_version' from 'torchmetrics.utilities.imports'
• 1.8 - not aligned dependencies:

   pytorch-lightning 1.8.0 depends on lightning-utilities==0.3.*
   torchmetrics 1.3.0.dev0 depends on lightning-utilities<0.10.0 and >=0.8.0
  

Before submitting

• Was this discussed/agreed via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure to update the docs?
• Did you write any new necessary tests?

PR review

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 🙃

---

📚 Documentation preview 📚: https://torchmetrics--2129.org.readthedocs.build/en/2129/

[codecov]

Codecov Report

Merging #2129 (42e6fef) into master (09bd064) will increase coverage by 0%.
The diff coverage is 100%.

Additional details and impacted files

@@          Coverage Diff           @@
##           master   #2129   +/-   ##
======================================
  Coverage      87%     87%           
======================================
  Files         293     293           
  Lines       16407   16415    +8     
======================================
+ Hits        14253   14261    +8     
  Misses       2154    2154           

[Borda]

@JustinGoheen @SkafteNicki seems the logging is different than expected for the latest PL (we have not tested it for long time 😿 ) mind have a look, pls: https://dev.azure.com/Lightning-AI/Metrics/_build/results?buildId=176883&view=logs&j=12a9e589-df1e-541c-f0d0-67a8d22922d8&t=b6423d8e-d265-5686-c4f1-df41b5bef40a&l=279

[Borda]

@awaelchli @justusschock, any ringing belts that could be the reason for the case failing...

[Borda]

Setting the upper bound for pl<2.0 for GPU issues but need to be investigated and unfreeze

• FAILED integrations/test_lightning.py::test_metric_lightning_log - AssertionError: assert -2.639833450317383 == tensor(-4.4507, device='cuda:0')

   >       assert epoch_0["metric_update_epoch"] == sum([model.sum[0], model.sum[1]])
   E       AssertionError: assert -2.639833450317383 == tensor(-4.4507, device='cuda:0')
   E        +  where tensor(-4.4507, device='cuda:0') = sum([tensor(3.3539, device='cuda:0'), tensor(-7.8046, device='cuda:0')])
  

• FAILED integrations/test_lightning.py::test_metric_collection_lightning_log - assert False

   >       assert torch.allclose(tensor(logged["SumMetric_epoch"]), model.sum, atol=2e-4)
   E       assert False
   E        +  where False = <built-in method allclose of type object at 0x7fc6260c6a40>(tensor(-22.7884), tensor(-18.4362), atol=0.0002)
   E        +    where <built-in method allclose of type object at 0x7fc6260c6a40> = torch.allclose
   E        +    and   tensor(-22.7884) = tensor(tensor(-22.7884))
   E        +    and   tensor(-18.4362) = TestModel(\n  (layer): Linear(in_features=32, out_features=2, bias=True)\n  (metric): MetricCollection(\n    (SumMetric): SumMetric()\n    (DiffMetric): DiffMetric()\n  )\n).sum