Bugfix/12167 pre: Figure label numbering for multiple figures not as subfigures

[kandolfp]

Description

As discussed in #12167 , the behavior for multiple top-level figures together with a label is not optimal. This might be a step in the direction of allowing this feature, at least it provides a warning if this combination is used.

It works well in my setup but of course I only have a very limited view on the possibilities for crossref, figures, subfigures, multiple figures, and layouts.

The main idea is to check if labelCellContainer === false but sortedOutputs is an array with more than 1 entry.
This allows to:

1. still define labels

   const outputLabel = label &&
        (
          labelCellContainer ||
          (Array.isArray(sortedOutputs) && (sortedOutputs.length > 1))
        ) &&
        isDisplayData(output)
        ? (label + "-" + nextOutputSuffix++)
        : label;

2. Emit a warning if a user combines multiple top-level figures with labels.

   if (labelCellContainer === false &&
        (Array.isArray(sortedOutputs) == true && (sortedOutputs.length > 1))
      ) {
        warning("Warning: using multiple top-level figures with labels might result in unwanted behaviour.")
      }

Downside, the variable naming labelCellContainer === false implies no more labels which is than contradicted.

Extended test cases

--- 
title: "foo"
keep-tex: true
---

Example form the [docs](https://quarto.org/docs/authoring/figures.html#layout):

```{python}
#| label: fig-test
#| layout-ncol: 2
#| fig-cap: 
#|   - "Line Plot 1"
#|   - "Line Plot 2"

import matplotlib.pyplot as plt
plt.plot([1,23,2,4])
plt.show()

plt.plot([8,65,23,90])
plt.show()
```

Referencing the different Figures is not working correctly

- `@fig-test`: @fig-test
- `@fig-test-1`: @fig-test-1
- `@fig-test-1`: @fig-test-2

```{python}
#| label: fig-test2
#| fig-cap: "multiple plots"
#| fig-subcap:
#|   - "Line Plot 1"
#|   - "Line Plot 2"

import matplotlib.pyplot as plt
plt.plot([1,23,2,4])
plt.show()

plt.plot([8,65,23,90])
plt.show()
```

- `@fig-test2`: @fig-test2
- `@fig-test2-1`: @fig-test2-1
- `@fig-test2-1`: @fig-test2-2

```{python}
#| label: fig-test3
#| fig-cap: "multiple plots"

import matplotlib.pyplot as plt
plt.plot([1,23,2,4])
plt.show()
```

- `@fig-test3`: @fig-test3

```{python}
#| label: fig-test4
#| fig-cap: "multiple plots"

import matplotlib.pyplot as plt
plt.plot([1,23,2,4])
plt.show()

plt.plot([8,65,23,90])
plt.show()
```

- `@fig-test4`: @fig-test4
- `@fig-test4-1`: @fig-test4-1
- `@fig-test4-1`: @fig-test4-2

Provides the expected output and warnings.

Checklist

I have (if applicable):

• filed a contributor agreement.
• referenced the GitHub issue this PR closes -> Figure numbering for multiple figures not as subfigures #12167
• updated the appropriate changelog in the PR
• ensured the present test suite passes
• added new tests
• created a separate documentation PR in Quarto's website repo and linked it to this PR

[kandolfp]

Obviously, my testing was not rigorose enough -> converted to a draft.