Optimizer kwargs should be passed as a dictionary rather than implicitly

[maurapintor]

The parameters passed to the run should be explicitly passed as optimizer kwargs incapsulated as a dictionary, rather than passing whichever additional input to the optimizer.

This also allow to extend the interface to pass other inputs to other parts of the run, e.g., the forward.

secml-torch/src/secmlt/adv/evasion/modular_attack.py

Lines 178 to 185 in 674b2bc

	def _run(
	self,
	model: BaseModel,
	samples: torch.Tensor,
	labels: torch.Tensor,
	init_deltas: torch.Tensor = None,
	**optim_kwargs,
	) -> tuple[torch.Tensor, torch.Tensor]:

Should be changed to

def _run(
        self,
        model: BaseModel,
        samples: torch.Tensor,
        labels: torch.Tensor,
        init_deltas: torch.Tensor = None,
        optim_kwargs: dict = dict(),  # pass as explicit dictionary
    ) -> tuple[torch.Tensor, torch.Tensor]:

[maurapintor]

Closed with #105.