[MNT] numpy 2 compatibility

[fkiraly]

Currently, the package is incompatible with numpy 2, though this is not restricted in the bounds.

If numpy 2 is installed, the failure traceback is

pytorch_forecasting/__init__.py:31: in <module>
    from pytorch_forecasting.models import (
pytorch_forecasting/models/__init__.py:4: in <module>
    from pytorch_forecasting.models.base_model import (
pytorch_forecasting/models/base_model.py:20: in <module>
    from numpy.lib.function_base import iterable
E   ModuleNotFoundError: No module named 'numpy.lib.function_base'

[fkiraly]

another failure, only on macos-13 this time:

Epoch 0:   0%|          | 0/33 [00:01<?, ?it/s]
________________________ test_prediction_with_dataframe ________________________

model = TemporalFusionTransformer(
  	"attention_head_size":               1
  	"categorical_groups":                {}
  	"ca...4,), eps=1e-05, elementwise_affine=True)
    )
  )
  (output_layer): Linear(in_features=4, out_features=1, bias=True)
)
data_with_covariates =          agency     sku    volume  ...     weight  time_idx  target
0     Agency_22  SKU_01   52.2720  ...   8.229938 ...1.963758        59   1.000
6772  Agency_22  SKU_04   72.0153  ...   9.486183        59   1.000

[180 rows x 31 columns]

    def test_prediction_with_dataframe(model, data_with_covariates):
>       model.predict(data_with_covariates, fast_dev_run=True)

tests/test_models/test_temporal_fusion_transformer.py:372: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
pytorch_forecasting/models/base_model.py:1432: in predict
    trainer.predict(self, dataloader)
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/trainer/trainer.py:858: in predict
    return call._call_and_handle_interrupt(
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/trainer/call.py:47: in _call_and_handle_interrupt
    return trainer_fn(*args, **kwargs)
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/trainer/trainer.py:897: in _predict_impl
    results = self._run(model, ckpt_path=ckpt_path)
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/trainer/trainer.py:981: in _run
    results = self._run_stage()
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/trainer/trainer.py:1020: in _run_stage
    return self.predict_loop.run()
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/loops/utilities.py:178: in _decorator
    return loop_run(self, *args, **kwargs)
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/loops/prediction_loop.py:121: in run
    batch, batch_idx, dataloader_idx = next(data_fetcher)
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/loops/fetchers.py:133: in __next__
    batch = super().__next__()
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/loops/fetchers.py:60: in __next__
    batch = next(self.iterator)
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/utilities/combined_loader.py:341: in __next__
    out = next(self._iterator)
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/lightning/pytorch/utilities/combined_loader.py:142: in __next__
    out = next(self.iterators[0])
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/torch/utils/data/dataloader.py:631: in __next__
    data = self._next_data()
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/torch/utils/data/dataloader.py:675: in _next_data
    data = self._dataset_fetcher.fetch(index)  # may raise StopIteration
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py:51: in fetch
    data = [self.dataset[idx] for idx in possibly_batched_index]
../../../hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py:51: in <listcomp>
    data = [self.dataset[idx] for idx in possibly_batched_index]
pytorch_forecasting/data/timeseries.py:1538: in __getitem__
    decoder_length = self.calculate_decoder_length(time[-1], sequence_length)
pytorch_forecasting/data/timeseries.py:1462: in calculate_decoder_length
    time_last - (self.min_prediction_idx - 1),
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

self = tensor(59), dtype = None

    def __array__(self, dtype=None):
        if has_torch_function_unary(self):
            return handle_torch_function(Tensor.__array__, (self,), self, dtype=dtype)
        if dtype is None:
>           return self.numpy()
E           RuntimeError: Numpy is not available