FEAT: add migrad_args option to Minuit2 optimizer

[redeboer]

Closes #475

[redeboer]

@Leongrim, could you try out the branch of this PR?

pip install git+https://github.com/redeboer/tensorwaves@migrad-arguments

optimizer = Minuit2(
    migrad_args=dict(\
        ncall=...,
        iterate=...,
    ),
)
optimizer.optimize(estimator, initial_parameters)

[Leongrim]

Works really well.
Only one thing. If the input is not an int the error Minuit throws can be a little confusing.

TypeError: __call__(): incompatible function arguments. The following argument types are supported:
    1. (self: iminuit._core.MnApplication, arg0: int, arg1: float) -> ROOT::Minuit2::FunctionMinimum

Invoked with: <iminuit._core.MnMigrad object at 0x7f62bc94a830>, 1000000.0, 0.1```

I'm not sure if this is something we should take into account or not.
But since it already changing ncall is something that should be rarely necessary to change.

[redeboer]

Only one thing. If the input is not an int the error Minuit throws can be a little confusing.

Which input type did you give? We could do some input evaluation (although that makes the code more complicated, possibly unnecessarily).

[redeboer]

Oh I see, nvm, looks like float.

[Leongrim]

I used a flaot (1e6).

[redeboer]

I would leave it like that -- no input evaluation on migrad_args. Seems like iminuit.Minuit2.migrad() documents its signature well, and the TensorWaves documentation on migrad_args refers to that page.