find_reversal_racetrack_filtered returning the original datapoint and not the bin datapoint

[Apelvd]

I believe the find_reversal_racetrack_filtered is returning the wrong vector.
It should return rev[ix[ixr]], ix[ixr] and not y[ix[ixr]], ix[ixr]

Here is the piece of code I was playing with to understand the behaviour of racetrack_filter:

y = np.random.normal(size=50)*10
fig = go.Figure()
time = x=[i for i, val in enumerate(y)]
fig.add_trace(go.Scatter(x=time,y=y, mode='markers', marker_size=10, name='Raw Data'))
binsDataFiltering = 6
reversal, indInit = fatpack.find_reversals(y, k=binsDataFiltering)
fig.add_trace((go.Scatter(x=indInit, y=reversal, name='find_reversal')))
reversal, indInit = fatpack.find_reversals_racetrack_filtered(y, h=0, k=binsDataFiltering)
fig.add_trace((go.Scatter(x=indInit, y=reversal, name='find_reversals_racetrack_filtered h=0')))


bound = fatpack.rainflow.get_load_class_boundaries(y, k=binsDataFiltering)
boundCouple = zip(bound[:-1], bound[1:])

for boundaryCouple in boundCouple:
    fig.add_hline(y=(boundaryCouple[1]+boundaryCouple[0])/2, line_dash='dash',  line_color='green')
fig.update_layout(title=dict(text="Rainflow counting with FatPack test"),xaxis=dict(title=dict(text="Time")),yaxis=dict(title=dict(text="physical value")))
fig.show()

Thanks for this great package !

[Gunnstein]

Dear @Apelvd,

I guess when comparing the behavior of find_reversal and find_reversal_racetrack_filtered one would expect that they both return the same thing, i.e. since find_reversal returns binned amplitudes, find_reversal_racetrack_filtered should also. (The behavior of find_reversal is defined in ISO 12110-2 and can therefore be adopted as "correct")

My only worry is that changing its behavior may introduce downstream issues for users, however, the difference is only significant in case k is small and I therefore dont think anyone "out there" will have a problem with changing to binned amplitudes.

Do you want to make a pull request with a fix of find_reversals_racetrack_filtered? Note that you may also need to change the results_true in the test case TestRaceTrackFilter.

In any case, thank you for the perceptive discovery and for reporting it!

Best regards,
Gunnstein