ENH: support more style params for guide_colorbar

[daylinmorgan]

This PR addresses #653, along with adding the other related parameters from the ggplot2 implementation.

Example:

import numpy as np
import pandas as pd

from plotnine import (
    aes,
    element_blank,
    geom_tile,
    ggplot,
    guide_colorbar,
    scale_fill_continuous,
    theme,
)

df = pd.DataFrame(
    {
        "value": np.random.randn(100),
        "variable": np.tile(["A", "B", "C", "D", "E"], 20),
        "time": np.repeat(np.arange(1, 21), 5),
    }
)

(
    ggplot(df, aes("variable", "time", fill="value"))
    + geom_tile()
    + scale_fill_continuous(
        guide=guide_colorbar(frame_color="black", ticks_color="black")
    )
    + theme(panel_background=element_blank(),figure_size=(3,3))
)

[has2k1]

@daylinmorgan I will look at this before of the next release. The styling needs to be integrated with the theming system. You can take a crack at it if you wish, otherwise I will do it.

[daylinmorgan]

The styling needs to be integrated with the theming system.

Could you elaborate more what you mean by this? Do these overlap with any of the current themeable elements?

[has2k1]

Could you elaborate more what you mean by this?

The rectangle for the colorbar should be themeable with element_rect(...) and, the lines for the ticks should be themeable with element_line(...).

How it works is, we store the Matplotlib artists that should be themeable in a dictionary theme._targets. Then a class in themes/themeable.py does the actual "theming" i.e. setting the properties of the artist.

[daylinmorgan]

Ah! It wasn't clear just using the docs for ggplot2 that the guide_colorbar actually supports a theme object as well.

[has2k1]

I have merged this in 18f397e4 and completely refactored the guides to mainly be styled through the local theme parameter or the global theme. Plus there is now full control over the positioning of guides.

[daylinmorgan]

Sounds great! Looking forward to the next release.