Use objectpath.Encoder to amortize the cost of encoding the paths of multiple objects

[yuxincs]

We are using objectpath.For very heavily in our inference engine to construct package-independent object identifiers. However, each invocation comes with a cost of initializations (construction and traversal of the scope tree).

Since x/tools v0.9.0 (we are currently on v0.11.0), there is an objectpath.Encoder that "amortizes the cost of encoding the paths of multiple objects". This PR uses the encoder for better performance. There aren't/shouldn't be any functionality changes.

A quick profile showing the timing before & after this PR for accumulation.run function (the entrypoint function for our inference engine):

before:

31.43s 22.49% |   go.uber.org/nilaway/accumulation.run
15.69s 49.92% |   go.uber.org/nilaway/inference.(*Engine).ObservePackage
8.11s 25.80%  |   go.uber.org/nilaway/inference.(*Engine).ObserveUpstream
3.33s 10.59%  |   go.uber.org/nilaway/inference.(*InferredMap).Export
...

after:

18.34s 14.63% |   go.uber.org/nilaway/accumulation.run
8.55s 46.62%  |   go.uber.org/nilaway/inference.(*Engine).ObserveUpstream
3.65s 19.90%  |   go.uber.org/nilaway/inference.(*InferredMap).Export
2.50s 13.63%  |   go.uber.org/nilaway/inference.(*Engine).ObservePackage
...

[sonalmahajan15]

Wow, this is quite promising! Let's confirm the benefits via performance evaluation internally, as well as, a sanity check for ensuring there are no functionality changes, i.e., the errors reported for a given set of test targets remain the same.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (f9441c8) 88.97% compared to head (336d11c) 88.97%.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #140   +/-   ##
=======================================
  Coverage   88.97%   88.97%           
=======================================
  Files          54       54           
  Lines        8253     8254    +1     
=======================================
+ Hits         7343     7344    +1     
  Misses        753      753           
  Partials      157      157           

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