the correct value seems to be 6389 for this gain setting etc. 2do:
double check that

though there is still a small display issue if the end of the display
buffer is hit because now only 1-2samples are associated with the pixel
and this causes the min/max and histogram displays to look odd.
shouldn’t be a big issue in real applications but worth fixing when we
inevitably refactor this whole plugin.

mostly to make space for amplifier saturation warning ui

still needs labels and better default values

Used to use spreadSelection->getSelectedId(), which wasn’t set properly
when the mouse wheel was used to change heights. Now goes directly for
the text value.

should also add a warning for actually saturating the amp- via
(configurable) thresholds

still needs more pointer checks for the pixel drawing when resizing -
can still crash fairly easily

and some smaller bug fixes

no more redundant redraws when scrolling

histogram method still needs _lots_ of optimization but at least its
not immediately crashing now.

Basically, we used to have many lfpChannelDisplays in the lfpDisplay,
each with its own repaint() calls. Now, because they all share one big
bitmap display, we need to clear a section of bitmap before we can draw
new data. This has to happen in lfpDisplay::Paint(), not in
lfpChannelDisplay::Paint() because channels overlap and doing it one at
a time would paint over already drawn data form the same buffer.
 Because the repaint() calls are asynchronous, this clearing in the
parent and then drawing the child component isn’t guaranteed to be in
sync - instead, the lfpChannelDisplay pixel-wise plotting is now done
in a new function, and the only redraw() is the one of lfpDisplay.

shift stops scroll wheel from reporting correctly on OS X, so channel
range is on ALT now instead

repaint is still wrong, missing vertical yellow line and left-right
borders are off by one buffer update

only replaced simple min-max code here, the supersampled one still uses
juice setPixel calls. This still doesn’t update the redraws correctly
and has all kinds of bugs but the basic functionality is there and the
cpu load is significantly lower in pretty much all cases.

event toggle buttons had inconsistent border widths

there’s still some fundamental rendering bug, the lines still have
brighter spots in places where it doesn’t quite make sense - maybe its
in the data though - best to make good synthetic data for testing all
the edge cases-  one sample steps in triangle waves, sines etc.

used to miss one sample/pixel, leaving small gaps occasionally

still needs significant performance optimization.
I’ll make a blog post about the details of why this approach should be
reasonable soon.

2do:
- debug, i’m pretty sure this is still missing one sample per pixel
somewhere, and the scaling form sample/histogram density to pixel color
is very hack-y
- performance
- make configurable via sliders