Implement to_float_tensor(values) that converts a Python list (or list of lists) of numbers into a float32 tinygrad Tensor.
The function should preserve the nesting structure of the input: a flat list becomes a 1-D tensor, and a list of lists becomes a 2-D tensor. The dtype must always be dtypes.float32, even when the input contains Python int values.
Tensor constructor accepts flat and nested Python lists; the shape is inferred from the nesting depth.dtype keyword. The float-32 value lives on the dtypes object you import from tinygrad.