Bump minimal dependencies

[d-a-bunin]

🚀 Task description

We could now update some of our dependencies that were blocked by old version of pytorch_forecasting

Plan

• Set python = ">=3.10.0, <3.13.0"
• Set torch = {version = "^2", optional = true}
• Set lightning = {version = "*", optional = true}
  • Remove pytorch_ligtning
• Set optuna = {version = "^3", optional = true}
  • pypi
  • Rework imports for everthing to work
  • Probably, sqlalchemy = {version = "^1.1.0", optional = true} could be removed
• Set statsforecast = {version = "^2", optional = true}
  • pypi
  • Remove etna.libs.statsforecast
• Set loguru = ">=0.7"
  • pypi
• Set catboost = "^1"
  • pypi
• Set numpy = "^2"
  • pypi
  • Updater numba accordingly
• Set pandas = "^2"
  • pypi
  • Rework pandas tests
• Set pmdarima = ">=2"
  • pypi
• Set hydra-slayer = ">=0.5"
  • pypi
• Set typer = ">=0.12
  • pypi
• Set holidays = ">=0.40
  • pypi
• Set wandb = {version = ">=0.16", optional = true}
• Set accelerate = {version = "^1", optional = true}
• Try to update poetry >= 2
  • Ensure to use poetry-plugin-export >= 1.9.0
• Make sure everything works fine and tests pass
• After bumping versions update pyproject.toml filterwarnings section
  • Remove redundant warnings
  • Add new warnings to ignore

To discuss:

• Should we try to bump optuna to 4.0? It was already release.
• Do we really want bump statsforecast to 2.0? It was released 2024-11-27.
• Do we really want to bump version of wandb?

Test cases

No response

Additional context

• Blocked by Remove pytorch-forecasting models #566
• Related to `DeepARModel` and `TFTModel` don't work on `pytorch_lightning>=2.0` #371
• Related to Update minimal python version up to 3.10 #420
• Related to Update minimal pandas version up to 2.0 #421
• Related to Add AutoTBATS and AutoMFLES from statsforecast #584
• Related to Update minimal numpy version up to 2.0 #597

[egoriyaa]

Earlier there were problems with huggingface-hub, so now version range is narrow. Try to widen the range after bump.

[d-a-bunin]

There are problems with updating numpy to >2:

• Numpy 2.0 support status? catboost/catboost#2671
• pmdarima is incompatible with numpy>=2.0.0 alkaline-ml/pmdarima#577