Improve title snipping in notebooks for revealjs

[cderv]

fixes #8012

Doing a title snipping when engine: jupyter is a bit too much for revealjs format because this will create a new title slide whereas when no title is provided, it should be considered custom first slide per our doc: https://quarto.org/docs/presentations/revealjs/index.html#title-slide

So I suggest we do not strip in Revealjs format.

Just asking for review because I don't fully understand why we do this stripping.

Does it sounds good ?

We also use the function for stripping at

quarto-cli/src/core/pandoc/pandoc-partition.ts

Lines 49 to 58 in 9f0c42f

	// partition markdown into yaml, the first heading, and the rest of the markdown text
	export function partitionMarkdown(markdown: string): PartitionedMarkdown {
	// partition out yaml
	const partitioned = partitionYamlFrontMatter(markdown);
	markdown = partitioned ? partitioned.markdown : markdown;
	// extract heading
	const { lines, headingText, headingAttr } = markdownWithExtractedHeading(
	markdown,
	);

So I need to check if we need also some support there... 🤔

it seems used for special cases like embed feature and some unused function

[dragonstyle]

I think the idea of this is that a python user with existing notebooks may want to render them using quarto render foo.ipynb --to html or similar commands and get a reasonable experience. Since it is common to place a heading as the 'title' of the notebook in a markdown cell at the top of the notebook, we're trying to find the best way to promote that to the document title in Quarto.

It's pretty tempting to just that if the user creates front matter at all, they should place a title there (not just for revealJS - just that once they're making front matter we expect them to handle the title)...

We could start with this if you're feeling conservative.

[cderv]

The revealjs specific case is that

• Using title will trigger the use of a specific title-slide sectionf from the revealjs template
• To opt out this title slide from the template, no providing title in YAML is for now the way to go.

Currently it seems you can't do that with engine: jupyter, or at least that we promote headers like ## without considering the slide-level, meaning that a non section slide could be promoted to section slide.

This is something quite specific to revealjs here related to this heading promoting feature. 🤔

[cderv]

BY the way the partitionMarkdown where we do use this splitting of heading, seems to have a bug regarding indexFile for caching. I'll report elsewhere for this

[cderv]

OK so it seems this is more complex that it seems. This markdown.headingText is used in a lot of places.

I am even not sure that the new addition in 24672a4 about not stripping header if content before heading (contentBeforeHeading) is really respected in the partitionMarkdown() function used in part of our code base

I'll leave this as draft for further work together on this, and open smaller PR for immediate issues found