Shift and add

[matteobachetti]

Basic implementation of the shift-and-add technique from Mendez+1998
Guide for review: the basic mechanism is in fourier.py, the examples in the doctests should be self-explanatory. InDynamicalCrossspectrum and DynamicalPowerspectrum, I just wrap the same functionality in convenient methods that unpack the relevant information from the dynamical spectra and use the function in fourier.py.

Example: if a QPO is changing its frequency during our observation

we can think of tracing its maximum (e.g. with dcs.trace_maximum(), here I did it by hand):

and then applying shift_and_add to improve signal-to-noise and, maybe, detect a second QPO:

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.11%. Comparing base (39431fc) to head (69bf6c5).
Report is 13 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #849      +/-   ##
==========================================
+ Coverage   96.08%   96.11%   +0.03%     
==========================================
  Files          48       48              
  Lines        9292     9371      +79     
==========================================
+ Hits         8928     9007      +79     
  Misses        364      364              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[matteolucchini1]

Other than the minor docstring comment, it is not immediately obvious to me (even from reading Mariano's paper) how exactly the new frequency axis is defined - e.g. in your example the new Frequency bins end up being [0.05 0.15 0.25 0.35 0.45] and I don't quite understand why. Maybe clarify that in the docstring (e.g. from "Array of output frequencies" to "array of output frequencies defined so and so").

Other than that, I think this is the kind of feature that should be documented explicitely in the dynamical powerspectrum section of the readthedocs.

[matteobachetti]

@matteolucchini1 I fixed the docstrings are requested. I also opened an Issue in the notebooks repository in order to add the tutorial in the the dynamical power spectrum notebook: StingraySoftware/notebooks#104

[matteolucchini1]

I have no further comments (other than the obvious fix of the broken link in the automatic doc checks). Before merging I would prefer if we could update the docs though. Can we think of a "standard" NICER/NuSTAR observation to get started?

[matteobachetti]

@matteolucchini1 there you go: StingraySoftware/notebooks#105
It was a good opportunity to refresh those tutorials!
I used an RXTE dataset, because this technique was used a lot in RXTE times and it was easier to find a good observation.

[matteolucchini1]

LGTM no further comments