Correct cpu_usage calculation when more than one signature

[heifner]

Change Description

• CPU usage of signature recovery was not being calculated correctly for multiple signatures. Correct calculation so that multiple signature recovery time is not overly calculated.

Consensus Changes

• Consensus Changes

API Changes

• API Changes

Documentation Additions

• Documentation Additions

[b1bart]

No need to block this PR as cpu billing is subjective however, this is a "change of policy" of sorts and we should consider documenting it somewhere or maintaining the old "policy"

[b1bart]

I assume this check had to go because the digest_time was not exactly equivalent and we were already verifying that cpu_time2 was >0?

Is there any value in maintaining the invariant that cache hits/misses bill exactly the same amount?

[heifner]

I don't see any value in maintaining it.

[heifner]

No need to block this PR as cpu billing is subjective however, this is a "change of policy" of sorts and we should consider documenting it somewhere or maintaining the old "policy"

It was not always this way. In 1.5.x it "correctly" calculated it. Although we have had other issues in actually using the correct calculation in the past. In short we have had a number of issues with correctly calculating and using the value. So much so that documentation of any given past policy would be difficult. For example: #6871.