Improve binary search for apply-load.

[dmkozh]

Description

The binary search uses t-statistic to ensure the necessary confidence at each step. This results in needing less samples far away from the true value and more samples around it. We still need at least 30 samples though to keep math simple. This works reasonably well, but still doesn't avoid some fundamental variance issues that we have that result in very different and statistically significant performance difference between different runs. That concern should hopefully be addressed separately.

Also fix some SAC max TPS benchmark issues, the math got messed up after I've updated it from 1s granularity to 50ms.

Checklist

• Reviewed the contributing document
• Rebased on top of master (no merge commits)
• Ran clang-format v8.0.0 (via make format or the Visual Studio extension)
• Compiles
• Ran all tests
• If change impacts performance, include supporting evidence per the performance document

[Copilot]

Pull request overview

This PR improves the binary search algorithm used in apply-load benchmarks by implementing a statistically-driven approach using t-statistics. The changes ensure more reliable performance measurements by adapting sample sizes based on confidence levels, while requiring a minimum of 30 samples for statistical validity. Additionally, it fixes issues with SAC max TPS benchmarking where the math was incorrectly handling the granularity change from 1s to 50ms.

Changes:

• Implemented new noisyBinarySearch function that uses t-statistics and adaptive sampling with configurable confidence levels
• Refactored benchmark functions to return single-ledger timing measurements for use with the new binary search algorithm
• Updated minimum ledger count requirement from >0 to ≥30 and removed the TARGET_CLOSE_TIME_STEP_MS validation constraint
• Added comprehensive test coverage for the noisy binary search algorithm with multiple variance and function types

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
src/simulation/ApplyLoad.cpp	Implements the new noisy binary search algorithm and refactors findMaxLimitsForModelTransaction and findMaxSacTps to use it
src/simulation/ApplyLoad.h	Adds function declarations for the new benchmark functions and exports noisyBinarySearch for testing
src/simulation/test/LoadGeneratorTests.cpp	Adds extensive test coverage for the noisy binary search with various scenarios and updates existing test configurations
src/main/Config.cpp	Removes the validation requiring APPLY_LOAD_TARGET_CLOSE_TIME_MS to be a multiple of TARGET_CLOSE_TIME_STEP_MS
src/main/CommandLine.cpp	Updates validation to require at least 30 ledgers and changes MIN_TPS comparison from >= to >
Builds/VisualStudio/stellar-core.vcxproj	Updates Visual Studio project file to include previously missing source files
Builds/VisualStudio/stellar-core.vcxproj.filters	Updates Visual Studio project filters to include previously missing source files