fix: strip squash-merge "(#N)" suffix from release-note bullets

[fiunchinho]

Summary

v1.3.1's release notes ended up with a duplicate PR reference:

- Tighten release-notes whitespace in cliff template (#10) in [#10](https://...) by [@fiunchinho](...)
                                                       ^^^^^    ^^^^^

The first (#10) is auto-appended by GitHub to squash-merge commit messages; the second is the explicit link we render via cliff's GitHub integration. The first becomes redundant once we have the second.

Adds a commit_preprocessor that strips the trailing (#N) from each commit message before parsing.

Diff

 sort_commits = "oldest"

+commit_preprocessors = [
+  { pattern = '(?m) \(#\d+\)$', replace = "" },
+]
+
 commit_parsers = [

(+ documentation comment explaining the (?m) flag is load-bearing.)

Why (?m)?

git-cliff's regex defaults to single-line mode, so $ matches end-of-text. The squash (#N) lives at end-of-subject-line, with the commit body following. Without (?m), the pattern silently fails to match. With it, $ matches end-of-line and the trailing ref strips cleanly. Bodies' (#N) references are unaffected.

Verified locally

$ git-cliff --tag-pattern 'v*.*.*' v1.3.0..v1.3.1 --strip all
### Fixed

- Tighten release-notes whitespace in cliff template in [#10](https://...) by [@fiunchinho](...)

Test plan

• Squash-merge this PR.
• Confirm v1.3.2 release renders without (#11) appearing twice in the bullet.