Treat early-stopped trials as pending

[ltiao]

Summary:

Context

We encounter duplicates in candidate generation when using an early-stopping strategy. So far this has been mitigated by specifying should_deduplicate=True in the GenerationNode. This appears to stem from an early-stopping trial f(x, t) stopped at progression t < T later being deemed promising by the surrogate when extrapolated to f(x, T) with fixed features.

Changes

This adds early-stopped trials to the pending observation features, which serves to prevent it from being suggested later.

TODO: Benchmarking

Differential Revision: D84842619

[meta-codesync[bot]]

@ltiao has exported this pull request. If you are a Meta employee, you can view the originating Diff in D84842619.

[codecov-commenter]

Codecov Report

:white_check_mark: All modified and coverable lines are covered by tests. :white_check_mark: Project coverage is 96.19%. Comparing base (2d1f813) to head (23cb330). :warning: Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4450   +/-   ##
=======================================
  Coverage   96.19%   96.19%           
=======================================
  Files         553      553           
  Lines       56687    56687           
=======================================
  Hits        54531    54531           
  Misses       2156     2156           

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

:rocket: New features to boost your workflow:

• :snowflake: Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• :package: JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.