pstmrk.it links evade autoamtic detection

Here is an example:

```
https://click.pstmrk.it/2sm/domain.ext%2Fpersonal-dashboard%3Futm_source%3Dsurvey_1_invite%26utm_medium%3Demail%26id=000/XyDGESkN/nSkn/FhpNV1Ok2s/c3VydmV5XzFfaW52aXRl
```
3 annoying things:
- there’s no scheme or `www`, with which we usually detect embedded domains
- the link is properly URL-encoded and stops at the first non-encoded `/` which probably works poorly with existing auto link extract. If I prefix the domain with `www.` to trigger detection and [run it](https://cimbali.github.io/CleanLinks/) the result looks like:
```
https://www.domain.ext/personal-dashboard?id=000%2FXyDGESkN%2FnSkn%2FFhpNV1Ok2s%2Fc3VydmV5XzFfaW52aXRl
```

Start by reproducing the provided pstmrk.it example and trace the automatic link-detection and URL-extraction entry points. The fix should recognize the encoded destination and produce the clean domain.ext URL without incorrectly stopping at the first unencoded slash; verify the result using the linked CleanLinks example.

Written by the indexing model from the issue text.