testing: bench: use --dataset instead of --size

[efiop]

This is no longer just about size (e.g. we have mnist/openorca/used-cars), so the --dataset makes more sense.

[codecov]

Codecov Report

Attention: 4 lines in your changes are missing coverage. Please review.

Comparison is base (c0b64b0) 90.60% compared to head (b9d6f6c) 90.37%.

Files	Patch %	Lines
dvc/testing/benchmarks/plugin.py	0.00%	3 Missing ⚠️
dvc/testing/benchmarks/fixtures.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #10134      +/-   ##
==========================================
- Coverage   90.60%   90.37%   -0.24%     
==========================================
  Files         494      494              
  Lines       37639    37639              
  Branches     5478     5478              
==========================================
- Hits        34104    34016      -88     
- Misses       2896     2963      +67     
- Partials      639      660      +21     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.