Remove type construction from tree_hash benchmarks

[macladson]

Issue

While doing some optimization experiments I noticed that we actually measure the type construction and the hashing in the same benchmarks which dilutes the apparent effect of any hashing optimizations.

Proposed Changes

Prevent the type construction from being timed by adding it into a "setup" closure which does not get benchmarked.
Doing it this way provides a "fresh" list every iteration so we avoid cache hits.

If we actually want to include benchmarks for construction we should make new benchmarks which specifically target that.

I also bumped criterion to 0.8

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 70.15%. Comparing base (4d88c72) to head (6e9cc61).

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #93   +/-   ##
=======================================
  Coverage   70.15%   70.15%           
=======================================
  Files          22       22           
  Lines        1280     1280           
=======================================
  Hits          898      898           
  Misses        382      382           

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[michaelsproul]

nice improvement!