Markdown [...](...) syntax does not play well with sphinx.ext.autosectionlabel
Describe the bug
The MyST documentation states that the [...](...) syntax is a superset of the any role. Yet, [...](...) does not appear to work for referring to section labels defined implicitly using sphinx.ext.autosectionlabel, whereas the any syntax does.
Reproduce the bug
index.md:
# Document title
Look at {ref}`the title <Document title>`.
Look at {any}`the title <Document title>`.
Look at [the title](Document title).
conf.py:
extensions = ["myst_parser", "sphinx.ext.autosectionlabel"]
Screenshot of output:

List your environment
MyST-parser master, Sphinx master. Also reproduced with Sphinx 4.5.
From further experimentation it looks like this problem shows up when the title contains spaces.
Oh, actually, I see now that markdown_it doesn't even recognize that syntax as a link. Reading the CommonMark spec, I see that I should have written [the title](<Document title>). Sorry for the wrong issue.
I do think this could use a note in the documentation.
I am starting to wonder if it would be appropriate for MyST not to require these brackets after all. For context, what triggered me to move my projects to MyST was this:
https://github.com/sphinx-contrib/sphinx-lint/issues/4#issuecomment-1109372243
Namely, quirks in reST linking syntax, which don't result in any warning ("errors should never pass silently"), and don't show the intended output. Now, I have to remember similar quirks with MyST as well :-(
So how about doing an extension to standard (CommonMark) Markdown to allow spaces and newlines in the target, even without pointy brackets <>?
So how about doing an extension to standard (CommonMark) Markdown to allow spaces and newlines in the target, even without pointy brackets <>?
Ah but that is standard commonmark, specifically: https://spec.commonmark.org/0.30/#example-487
This is not really an extension to commonmark, it would be effectively breaking it. So I'm afraid even it is possible, it's highly unlikely to happen
OK. Too bad, but I understand.
Some possibilities remain not explored, though, e.g. this sort of mix between reST role with target syntax and Markdown doesn't look totally unconceivable to me:
{ref}[the title](Document title)
or (in a fashion less similar to CommonMark, so it shouldn't classify as "breaking" the spec):
{ref}`The title`<Document title>
Or something. At any rate, like the folks in #548, I wish MyST had a brainless way of doing cross-references. The usual problem with {ref}`Text <target>` is that there must be a space before the <, which everyone forgets about…