Refactoring scipy solver options into their own pydantic models

[gialmisi]

The options being passed to scipy solvers here and here should be refactored into their own pydantic classes, as is done for the Nevergrad solvers here. In other words, the initializers of ScipyDeSolver and ScipyMinimizeSolver should have the following structure:

def __init__(self, problem: Problem, options: ScipyDeOptions):
    ...

def __init__(self, problem: Problem, options: ScipyMinimizeOptions):
    ...

where ScipyDeOptions and ScipyMinimizeOptions should be instances of new pydantic classes. For guidance on how these should look like, see here. Default options should also be defined as their own instances of these new classes (such as is done here).

Writing also a couple tests more for the two scipy solvers would not hurt.