Ask for help: Questions about the number of runs and multiple threads of the retesting system policy

The platform building. Background: I wanted to use machine learning to optimize the parameters of the strategy, instead of using the customization functionality of our platform, because our platform's violent parameter optimization method is less efficient when targeting a large number of parameters.

The problem is: 1. When retesting, the logic of my program is to have my policy run 100 times automatically every time inside my program, and then search for the optimal parameter. But when we retest the platform, it seems that clicking the retest button only stops automatically every time it runs once, please ask how to solve this? 2. Does our platform support a set of strategies to open two threads, then one thread to retest optimized parameters based on real-time market, one thread to real-time trading, retest optimized parameters can be transmitted to real-time trading. (If not, then can you trade through the testnet of the exchange, so as to achieve the retesting purpose?