As is stated in the cargo-difftests README, one could simply use cargo-difftests like so:
Recommended workflow
Initial profiling data collection
Analyze, rerun and collect new profiling data
cargo difftests rerun-dirty-from-indexes is basically an alias for cargo difftests analyze-all-from-index --action=rerun-dirty; more information about analyze-all can be found in the analyze-all article.
The --compile-index tells cargo-difftests to compile indexes. You can optionally pass --and-clean next to --compile-index to have cargo-difftests clean up all the profiling data other than the index after it’s done running a test, to reduce the amount of required disk space.