docs: Add example of cumulative line chart with facet

[dsmedia]

Description

This PR replaces the "Cumulative Wikipedia Donations" example with a new chart, based on issue #2424. The original example was broken and went against the Contribution guideline that requires local datasets, and a new faceted cumulative sum chart was proposed in 2021. This PR incorporates the suggestion made by @jakevdp to improve x-axis labels. The example was developed and proposed by @harabat, but had not been 100% finalized. I made some very small tweaks to get it working in Altair v5 and to incorporate small recommendations in #2424.

Changes Made

• Introduced a new cumulative sum chart to replace the broken (and since-removed) example of Cumulative Wikipedia Donations.

The following code provides an illustration of a cumulative sum (to replace the broken chart), but also clarifies the styling of titles, headers, order, and marks of a facet plot, and gives an example of combining predicates with the filter transform.

Source: @harabat

• Improved x-axis label formatting per @jakevdp suggestion.
• Assigned to category of line charts. (@jakevdp in 2021 suggested this be categorized as a case study, but the example gallery has evolved significantly since then. This example builds on a simpler cumulative line example and seemed to be a good fit within that section.)

Issue Reference

Resolves #2424

[joelostblom]

Thanks for picking this up @dsmedia! I made a few suggestions to improve the original example here. I also think this might be a better fit for the advanced calculation section in the gallery since as you noted there is already a cumulative chart in the line category and the most interesting thing here is probably the more intricate use of filters.

[mattijn]

Ignore if you had a specific intend with the current scales, but I noticed there is a slight misalignment, especially the x-axis of 'epidemic' and 'flood'. This could be resolved by changing

.resolve_scale(y='independent')

to:

.resolve_axis(y='independent', x='independent')

Where the result will look like this:

But again, if the current approach was with a certain intend, then please ignore this comment.

---

Also linking #695. Maybe someone sees this as an opportunity to improve this current syntax

.transform_filter(
    {'and': [
        alt.FieldOneOfPredicate(field='Entity', oneOf=columns_sorted), # Filter data to show only disasters in columns_sorted
        alt.FieldRangePredicate(field='Year', range=[1900, 2000]) # Filter data to show only 20th century
    ]}
)

Since this:

alt.FieldOneOfPredicate(field='Entity', oneOf=columns_sorted) & alt.FieldRangePredicate(field='Year', range=[1900, 2000])

Returns currently:

TypeError: unsupported operand type(s) for &: 'FieldOneOfPredicate' and 'FieldRangePredicate'

[joelostblom]

I like the change suggested by @mattijn to make the axes clearer. If you make that change @dsmedia we can go ahead and merge after.

[dsmedia]

Thanks for the suggestion, @mattijn. I've submitted the change.

[mattijn]

Thanks looks good!