fix(ci): ensure deploy workflow does not override benchmark data

[pichlermarc]

Which problem is this PR solving?

Currently out docs deploy workflow deletes the pre-existing benchmark data from the benchmarks/ directory from the gh-pages branch (see 6cf6b67, b3a3d45).

To fix this, this PR:

• adds the clean-exclude option with benchmarks/** to ensure they are not deleted when the other docs data is rsynced by the action.
• updates the deploy action to v4 (at least v4.2.0 is needed for wildcard matching in clean-exclude)
• updates the node version used in the workflow to v20 (so that we can use v4 of the deploy action)

How Has This Been Tested?

• Ran the workflow in a private repo where I pushed both the main and gh-pages branch.

[codecov]

Codecov Report

Merging #4376 (cf45bdd) into main (3e59291) will not change coverage.
The diff coverage is n/a.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4376   +/-   ##
=======================================
  Coverage   92.22%   92.22%           
=======================================
  Files         332      332           
  Lines        9467     9467           
  Branches     2011     2011           
=======================================
  Hits         8731     8731           
  Misses        736      736           

[pichlermarc]

we don't need this anymore as the action uses the default authentication since v4.