revert: import remark-gfm library in NcRichText as module, not async

[Antreesy]

☑️ Resolves

• This reverts commit 49ce673. (fix(NcRichText): async import remark-gfm library #6259)

After some testing, I'm not happy with the results. async load of ~40kB comes with side effect of resizing of components, which doesn't look good in long lists (Talk)

🖼️ Screenshots

🏚️ Before

2025-02-11_10h36_17.mp4

🏡 After

2025-02-11_10h44_40.mp4

🏁 Checklist

• ⛑️ Tests are included or are not applicable
• 📘 Component documentation has been extended, updated or is not applicable
• 3️⃣ Backport to next requested with a Vue 3 upgrade

[susnux]

Would it not be possible to just keep showing the skeletons until loading is finished?
That would fix the visual issue and also keep bundle size smaller (also performance of parsing).

[ShGKme]

Would it not be possible to just keep showing the skeletons until loading is finished?
That would fix the visual issue and also keep the bundle size smaller (also performance of parsing).

We could show the text until the (I missed that it shows only a spinner). But even with the text, we cannot predict the result size, having all the headings, lists, code blocks and so on.

[ShGKme]

Agree, I wasn't fun of it initially, as mentioned in the original PR

[Antreesy]

Would it not be possible to just keep showing the skeletons until loading is finished?

If you're asking whether it's possible to do on app side, we then need to trigger async import while it's loading (e.g. place NcRichText with markdown somewhere hidden to trigger it on initial load), and show skeleton until internal component state is updated? Not sure that's a good idea..

[ShGKme]

also keep bundle size smaller

We can separate it into two components:

1. Simple and light text renderer with mentions and links support
2. Heavy markdown renderer

[susnux]

It was more like:
emit some loaded event when rendering is finished so the app can replace their loading (talk skeletons) with the rendered component.

We can separate it into two components:

Do we use the non-markdown versions somewhere? Because if not then it makes not much sense.
For me the problem is the amount and size of dependencies that are synchronously imported by this component.
So for personal projects I do not use it but custom components with markdown rendering saving quite some size.

[susnux]

anyways makes sense here

[Antreesy]

Do we use the non-markdown versions somewhere?

Found some usages in Talk, Activity, integration apps, could be somewhere else. So some may benefit from it, especially if discussed strip-markdown option will be added

[Antreesy]

/backport to next