Fix probems for fitting

[andrea-pasquale]

In this PR I've addressed a few issues that were raised in the last meeting.

1. If the fit fails it will not raise an error but just a warning so that the routines pipeline can continue regardless of the output of the fitting.
2. Also the problems related to the parameter points are fixed now thanks to (1).
3. I've also tried to lower the number of warnings generated by qq-live.
4. Fix a small bug related to whether the lorentzian fit should find a min or a max depending on the routine and on the platform.

Finally, in the case where we have both a fast sweep and a precision sweep, the fitting before acted separately first on the fast sweep data and then on the precision sweep data. In order to have more statistics I've modified the fitting such that when the precision sweep starts the fitting will not discard the fast sweep data but it will use both datasets to perform the fitting.
In this way we can avoid some weird results when we start generating the precision sweep data.
Let me know if you like this approach @aorgazf @DavidSarlle @maxhant.

[maxhant]

@andrea-pasquale I haven't tested it but I guess that it will print fitting unsuccessful on the first few points right?

[andrea-pasquale]

@andrea-pasquale I haven't tested it but I guess that it will print fitting unsuccessful on the first few points right?

Exactly @maxhant.
https://github.com/qiboteam/qcvv/blob/fd1b914eabda8c20592be82e1bc5759baf8d16b4/src/qcvv/fitting/methods.py#L68

[DavidSarlle]

@andrea-pasquale I have includede the changes in PR #56 #57 and it works for me.