The short answer is no, QIIME2 does not have a built-in method for doing this. Use Bray-Curtis for genus-level beta diversity analyses.
UniFrac with genus-level information does not really make much sense because the method is intended to calculate sample similarity based on phylogenetic distance (though distance on any tree could be used, I suppose). With genus-level information, some classifications may be wrong, ambiguous (causing potentially distant clades to be collapsed into a single label!) and/or not represent true phylogenetic structure, and in any case it would be a reduction in the level of information present in the individual ASVs/OTUs themselves.
Asking whether that reduction in information still produces the same patterns is an important and useful question. The easiest way is to use Bray-Curtis on the collapsed table.
The second way is to use q2-vsearch to perform de novo OTU clustering on your table at different similarity thresholds. This would output a new table and new sequences for building a collapsed phylogenetic tree for UniFrac. This analysis is not based on taxonomy, but on sequence similarity, which may be more sensitive for this type of analysis (since again, taxonomy can be wrong or arbitrary). Does that make sense?
I hope that helps! I am interested to hear your thoughts and the overall goal of your analysis.