Improved the capabilities of add_gradient_fill

[cedrichol]

add_gradient_fill() now supports:

• log plots (solves issue #17)
• having a tuple(minimum, maximum) alpha instead of a single float (which is still supported)
• Starting the gradient at:
  • the bottom of the plot (like the underglow, solves issue #16)
  • the minimum of the line (previous behaviour)
  • the top of the plot
  • the max of the line
  • zero (the visual effect is really nice depending on the signification of the data...)
• choosing the number of sampling points

Other improvements :

• internally imshow() is called with [interpolation='bilinear'], which helps reduce the necessary sampling points for linear plots
• xlims and ylims are now saved and restored, as add_underglow() does, for the same reasons

test_plotting_gradient() shows the capabilities of this implementation:

[diceroll123]

My hopes were risen and dashed when I thought this fixed #15! 😭 EDIT: This has inspired me to look into #15 more. I've updated #15 with some findings, it may be doable, however a large iterative project if mplcyberpunk were to work for a larger assortment of plot types! 🤔

That aside, looks amazing. Great work!

[cedrichol]

Thanks !

As for #15, I might take a look at it, it might be doable

[dhaitz]

wow, this looks really awesome, especially with the docstring, tests and detailed PR! Thanks for your contribution. I've made some minor edits (mostly naming), should be fine now!

[diceroll123]

I've closed #15 and #16 as completed now.

#17 can be closed, me thinks.

[cedrichol]

Happy you like it, thanks for this cool project btw