Added missing slices of data samples arrays in spectrum calc

[demvlad]

Resolved issue of missing slices of data samples arrays in spectrum calc.
The FFT used huge rezerved data length input before.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[nerdCopter]

we dont have Netlify previews at all anymore @blckmn ?

[Copilot]

Pull Request Overview

This PR fixes an issue with the FFT calculation by correcting the sample slicing for spectrum processing. Key changes include:

• Truncating the samples array in _getFlightSamplesFreq to use only populated data.
• Applying slicing to vsValues arrays in _getFlightSamplesFreqVsX.
• Slicing the piderror and setpoint arrays in _getFlightSamplesPidErrorVsSetpoint to match the actual sample count.

Comments suppressed due to low confidence (3)

src/graph_spectrum_calc.js:302

• Confirm that slicing using samplesCount does not exceed the allocated array length and that samplesCount is properly maintained.

    samples : samples.slice(0, samplesCount),

src/graph_spectrum_calc.js:418

• Verify that slicing the 'piderror' TypedArray to samplesCount preserves the intended data and aligns with downstream FFT processing.

    piderror: piderror.slice(0, samplesCount),

src/graph_spectrum_calc.js:419

• Ensure the sliced 'setpoint' array correctly reflects the populated data count and remains in sync with the corresponding 'piderror' values.

    setpoint: setpoint.slice(0, samplesCount),

[nerdCopter]

• approving untested.

[demvlad]

It was not big issue as i've thought on begin :). The FFT get real data count as parameter. But it allocate float data memory for huge data count, what is not good.

[nerdCopter]

i should not have approved untested.

several reports of breakage.
tested master, this PR has broken "spectrum graphs"

[haslinghuis]

@demvlad please have a look at #828

[demvlad]

@haslinghuis @nerdCopter
I doubted that it was bug and i've found an answer.
The source spectrum code used FFT zero padding method, to get bigger fft resolution at the chart. It uses big input/output size with big zero values pad. As result we had high fft resolution as fs/N. But zero pad does not change frequency resolution, what is 1/T.
I want try to use fft input size as nearest power 2 value and fill zero values in empty cells only.
The both methods is working. But we need to test and compare new method before.
Those methods choice is ground of my next PSD PRs. I will need the big rebase of PSD PRs in any case. I will use new method for PSD.
Therefore i suggest to revert this and PSD :) PRs.
The PSD works good and i have power2 sized version, but let's test and compare the all these before.