Refine the README.md of "Porting Benchmark"

[xuafeng]

Based on Bohan's experience, the doc (https://github.com/sys-intelligence/system-intelligence-benchmark/blob/main/doc/porting_benchmark.md) is still not very clear. We should tell which parts are needed, and which parts are optional. Highlight the most important steps (gitsubtree; model_config; implement main.py - if you already decouple executor/evaluator, you can simply add wrapper here to expose model/agent as parameters; install.sh, run.sh, README.md), make others as optional.

[xuafeng]

Resolved #32