For derendering latex in Emacs, see https://github.com/io12/org-fragtog.
For drawing images in line, you could try https://github.com/misohena/el-easydraw.
With Obsidian, I think you can get the Excalidraw plugin to draw images, though it's not inline (it opens a new pane).
Not a perfect match for your requirements, but with some features that might meet some of the same underlying goals, you might want to take a look at Zettlr.
This does not have inline images, right? This is basically the most important feature, that is missing from emacs org-mode.
Does not match your requirements exactly, but may come close (source and auto rebuilt PDF side-by-side with synctex to jump from one to another, etc) - VSCode with LaTeX Workshop extension. Not sure what VSCode can do for inline editing of images - but perhaps there is an extension that can help. It is capable of collapsing/expanding sections/subsections in the source view, and I think it does have a tree-style outline editing in a side-pannel too.
Is there a text editor that allows me to
enter normal text,
latex that is rendered in line, but is derendered when my cursor is over the latex,
draw images in line directly in the text editor, e.g. using a drawing tablet,
has headings like emacs org mode, that you can collapse the headings and nest them arbitrarily deep, and you can rearrange the headings. For example if you think of the headings as trees, we can move an entire subtree of headings at a certain level of depth before or after leaf notes in the tree, e.g. that exists a shortcut, such that I can turn this tree
item 1 -item 2
item 3
into this one
item 3
item 1 -item 2
Ideally the program would also have a mobile app, where you can edit your notes from.
vim mode would be nice too
being a zettlekasten like roam research would be nice
Here a program so I already know which don't satisfy all of the requirements. Especially the inline image drawing one.
So far nothing seems better than org mode.
After using obsidian for 2 years I switched back to Emacs. I made it such that I can use typst. I preview documents in the browser, so my documents can in principle contain anything that the browser can display (or run!).
To figure out things I first write things on a whiteboard, before writing down text.