change: Add validation decorators for core components

[mathysgrapotte]

Is your change request related to a problem? Please describe.

Data type mismatches between pipeline stages cause cryptic failures late in execution.

Describe the solution you'd like

1. Implement @validate_io decorator in utils/validation.py
2. Decorate critical methods in encoders/splitters/loaders
3. Add type checks for inputs/outputs with context-aware errors

[mathysgrapotte]

example implementation for context :

from functools import wraps
from typing import Type, Tuple, Optional
from stimulus.exceptions import StimulusError

def validate_input_types(*types: Type[object]):
    def decorator(func):
        @wraps(func)
        def wrapper(*args, **kwargs):
            for i, (arg, expected_type) in enumerate(zip(args[1:], types)):  # Skip self
                if not isinstance(arg, expected_type):
                    raise StimulusError(
                        f"Invalid type for argument {i} in {func.__name__}"
                    ).add_context(
                        expected_type=expected_type.__name__,
                        actual_type=type(arg).__name__,
                        argument_index=i
                    )
            return func(*args, **kwargs)
        return wrapper
    return decorator

# Usage in encoder
class TextEncoder:
    @validate_input_types(str)
    def encode(self, text: str) -> torch.Tensor:
        return self._encoding_logic(text)