Add automated workflow for pid

[nirmitparikh8]

fixes #6619

Why This Is Important

This PR establishes automated validation for the adaptive PID controller circuit breaker implementation, ensuring performance regression detection and providing continuous benchmarking against classic circuit breakers across multiple failure scenarios.

What We're Accomplishing

• Continuous Integration: Testing of PID controller and classic circuit breaker performance on every PR
• Performance Validation: Detection of regressions in circuit breaker behavior
• Visual Documentation: Ensuring performance graphs are updated with every commit

Key Changes Made

1. CI Workflow Setup

• Created .github/workflows/automated-experiment-result-checkeryml
• Check if latest commit updated all the graphs that indicate true performance of teh circuit breaker. If any images are outdated CI fails and tells the user what command to run.

2. Experiment Infrastructure Refactor

• Reorganized experiment structure: Moved test files to experiments/experiments/ directory and results into experiments/results/ directory
• Parallel test execution: Created run_all_experiments.rb with threaded execution (14 concurrent tests)
• Automated graph generation (requests, duration, throughput)

3. Test Coverage

14 test scenarios comparing adaptive vs classic circuit breakers:

• Sustained load (20% error rate)
• Error spikes (20% and 100% rates)
• Gradual error increases
• Oscillating error patterns
• Multiple service degradation scenarios

NOTE: We skip the lower_bound_adaptive test for the PID controller because it takes an hour. We will create a ticket to make it shorter or more optimized and then include it.

Future Steps

• Block merging to Semian main branch if all CI checks don't pass.
• Set something up where we can comment on the PR to generate updated graphs and commit back to the PR

This new check will automatically help catch any regressions introduced by code changes.

[AbdulRahmanAlHamali]

I added a bunch of comments. But looks good overall!

A couple of thoughts:

1. It seems CI did not run on your latest commit. I wonder if the last commit being made by the github action bot s blocking CI. Could you confirm?
2. Also, we probably wanna do a fast followup after this that adds a CI that basically confirms that all the experiments have been rerun (basically by confirming that every PR is updating all existing results).