enable infer api with multi-threads

[tensor-tang]

According to @Superjomn 's suggestion, the usage should be like this:

  auto main_predictor = CreatePaddlePredictor(config);

  std::vector<std::thread> threads;
  for (size_t i = 0; i < num_threads; ++i) {
    threads.emplace_back([&, i](
      // clone a predictor which is thread-safe
      // and shared the same parameters
      auto predictor = main_predictor->Clone();
      // prepare inputs and output ...
      predictor->Run(inputs, outputs);
    ));
  }
  // join...

So, we should only focus on thread-safe and sharing parameters.

[panyx0718]

Can this be unique_ptr? and std::moved around?

[panyx0718]

Why is there no change in CreatePaddlePredictor?

[tensor-tang]

Can this be unique_ptr? and std::moved around?

I am afraid not, this scope should be parent scope and the sub scope should be depend on it

Why is there no change in CreatePaddlePredictor?

Because as default only need change Clone.

[panyx0718]

make it explicit?

[tensor-tang]

Sure~ will update it.