Don't use BrailleCanvas for vertical histogram by default

[topolarity]

The vertical histogram uses Unicode Block Elements (U+2580–U+259F) for most of its diagram, but it inherits its "blank space" Braille (U+2800) character from the active canvas.

In the terminal this works fine since everything is monospace, but in browsers this often causes visual mis-alignment since, e.g., on macOS the default fonts give Braille characters a different spacing.

Before:

       ┌          ┐
   130  █⠀⠀⠀⠀⠀⠀⠀⠀▇
        █⠀⠀⠀⠀⠀⠀⠀⠀█
        █⠀⠀⠀⠀⠀⠀⠀⠀█
        █⠀⠀⠀⠀⠀⠀⠀⠀█
        █⠀⠀⠀⠀⠀⠀⠀⠀█
        █⠀⠀⠀⠀⠀⠀⠀⠀█
        █⠀⠀⠀⠀⠀⠀⠀⠀█
        █⠀⠀⠀⠀⠀⠀⠀⠀█
        █▄⠀⠀⠀⠀⠀⠀▄█
        ██▂⠀⠀⠀⠀▃██
        ███▇▆▅▇███
        ██████████
        ██████████
        ██████████
     0  ██████████
       └          ┘
       ⠀0⠀⠀⠀⠀⠀⠀⠀⠀2⠀
       μ ± σ: 1.0 ± 0.71

After:

       ┌          ┐
   130  █        ▇
        █        █
        █        █
        █        █
        █        █
        █        █
        █        █
        █        █
        █▄      ▄█
        ██▂    ▃██
        ███▇▆▅▇███
        ██████████
        ██████████
        ██████████
     0  ██████████
       └          ┘
        0        2
       μ ± σ: 1.0 ± 0.71

The remaining mis-alignment issues in the browser are due to Github, Slack, etc. using a font-family that causes some glyphs to be filled in by one font and others to be provided by another. We can't do much about that, but at least we can fix this

[topolarity]

pre failures are because the hash implementation in Julia 1.13, and Contours.jl seeds its algorithm with the first(...) entry of a dict (the contour lines changed, so the plot did as well)

Not related to this PR anyway

[t-bltg]

Oh I see you are right, vertical histogram have slightly diverged from the original implementation of horizontal histograms which are using BlockCanvases.

I have to cleanup the implementation at some point but your fix is very much valid.