[native_assets_builder] Benchmark hook overheads

[dcharkes]

When we want to remove the experimental flag for hooks we should check what the typical overhead is for hooks:

1. If there are no hooks at all in the package graph, do dart and flutter commands not regress.
2. The first-time hooks are run, what is the overhead of the hook compared to the actual work done in the hook. (Actual work: building/downloading a binary.)
3. For the 2nd-time run, how much is the overhead of the md5 file hash checking.