Crash (assert failure in stl operator[] '__glibcxx_requires_subscript') in GLScope::drawHistogramChannelGraph Histogram mode when switching between T-Y and X-Y view a few times

[ericfont]

Describe the bug Crash (assert failure in stl operator[] __glibcxx_requires_subscript) in GLScope::drawHistogramChannelGraph Histogram mode when switching between T-Y and X-Y view a few times.

To Reproduce Steps to reproduce the behavior:

1. Using this config file which is basically just CH2 enabled at 2V scale and timebase 5 us and Samplerate 12 MS/s and calibration out of 100 kHz and HISTOGRAM Mode enabled: before-crash.conf.zip
2. Select Horizontal format T-Y
3. Select Horizontal format Y-X
4. If doesn't crash yet, then keep switching between Y-X mode and T-Y mode until crash...crash always seems to happen on the switch from Y-X to T-Y and but doesn't happen when switching from T-Y to Y-X.
5. Assert crash is raised with the following stack trace (screenshot from QtCreator):

note that channel parameter is 1, however the graph.vaoHistogram array is empty. So it tries to access an array in memory which it shouldn't be allowed to.

Expected behavior Should be able to switch between X-Y and Y-X mode if the program allows me too. Either that or program should prohibit histogram mode or disable it when switching.

Screenshots If applicable, add screenshots or a video to help explain your problem.

Computer environment (please complete the following information):

• OpenHantek version: [use always the latest commit, e.g. release v2.03, commit 1a2b3c4, ...]
• OS: linux
• Distribution, version: Ubuntu 20.04 LTS (KDE Neon up-to-date)
• Video hardware: Radeon
• Qt version: Qt 5.12
• C++ compiler version: gcc 9.3.0

Scope device (please complete the following information):

• Device *Hantek 6022BL
• Program top line OpenHantek6022 self compile (20211118 - commit 927 (c55fe595cee5ef50c5dafda942e7ca7c50fb3797) - Device DSO-6022BL (FW0209)
• Input signal: 100 kHz Calibration out square wave
• Probe setting: both probes on 1X. Probe 2 is connected to calibrartion. Only CH2 enabled in OpenHantek.

Additional context This bug happens on main branch from before my last PR, so my PR wasn't at fault.

[ericfont]

My initial theory is that upon switching, the Histogram isn't immediately created internally, and so then if a GLScope::paintEvent is triggered by the OS before the Histogram is created, then there is a risk that line 550 of glscope.cpp is called to drawHistogramChannelGraph, since the scope->histogram condition is true even though the vaoHistogram vector is empty.

[ericfont]

See https://github.com/OpenHantek/OpenHantek6022/pull/241 for a simple fix.