Build iOS example app for metal MSM benchmark

[moven0831]

• close Update gpu-acceleration-app with latest mopro v0.2.0 integration #74

with initial benchmark on iOS as below

2^14: 319 ms
2^15: 431 ms
2^16: 512 ms
2^17: 654 ms
2^18: 915 ms
2^19: 1335 ms
2^20: 2158 ms
2^21: 3690 ms
2^22: 6855 ms

compared with our v1 MSM implementation (a553c49 and docs here: https://hackmd.io/@FoodChain/SJRjE0Nh6), this shows significant improvement on

1. better timing (e.g. for 2^20 test case, v1: 41019 ms; v2: 2158 ms. ~20x improvement)
2. better resource management (e.g. memory)
3. better GPU usage of ALU/f32 uints