Non-DOI links to `https://doi.org` trigger obscure error

[agoose77]

Description

A simple link to https://doi.org, e.g. [DOI](http://www.doi.org/index.html) causes the MyST build to crash with an obscure

This format is not supported or recognized

error.

Proposed solution

Only treat the link as a DOI if it matches r"https:\/\/doi\.org\/10\.\d+\/.*"

[rowanc1]

One thing to keep in mind is that there is the short-doi service, where dois don't follow the normal rules but are still valid:

For example:
https://doi.org/cr3qwn
https://doi.org/10.1175/1520-0493(1972)100<0081:OTAOSH>2.3.CO;2 (which is a terrible URL, but is real)

This was introduced here where we now no longer validate a DOI URL after it is recognized:
#735

I think we can treat this as a new error, and treat non-resolving DOIs in the transform a bit better.

[rowanc1]

No longer crashes, and has a better error message:

⛔️ page.md:63 Citation data from doi.org was not available or malformed for doi:index.html
   To resolve this error, visit https://doi.org/index.html and add citation info to local BibTeX file

[mfisher87]

Encountered this issue today. The error message isn't bad:

For others who might encounter this and use strict builds, I'm working around it with the following in my myst.yml, under the project key:

  error_rules:
    - rule: "doi-link-valid"
      severity: "warn"

I was not able to find a key value for this error rule that worked, so this is downgrading all checks that DOI links are valid.

What about instead of resolving doi.org links as DOI references, using the @ citation syntax with a prefix? E.g. @doi:10.5281/zenodo.17716243