Add weights support to qffa search

[YaoYLastro]

• Added a weights parameter to the folding algorithms for handling light curve intensities.
• Modified functions search_with_qffa and search_with_qffa_step to support weights input.

[matteobachetti]

@YaoYLastro thanks for your contribution!
I'd like you to write a test showing that the algorithm works as expected, e.g. creating a time series with a pulsation that gets recovered correctly by the algorithm.
I fear that, in the current form, the algorithm with have a lot of side effects if you intend the weights as photon counts, probably not as much if you consider weights à la Fermi (e.g. probability that a given photon is from the source).

[matteobachetti]

If I'm not mistaken, the algorithm should work even if you keep the weight keyword to None. If you force it to be np.ones, you add an additional multiplication in all steps that most of the time will be unneeded, making the algorithm a lot less efficient in the typical use case.