Support dynamic calculations on trained model

We should be able to support different kinds of dynamic calculations on the trained model.
At the moment some of these operations are exposed as methods on the model found inside the result object.

One example is the exploration of the model's representation of a single point in https://github.com/novonordisk-research/ProcessOptimizer/blob/develop/examples/7Dimensional_optimization_and_plotting.ipynb where the following code is executed to generate further data
```python
samples_of_y = res.models[-1].sample_y(transformed_point, n_samples=25000, random_state=None)`
```
These calculations has to be made on the server as it is the only place we have access to python. The solution currently pursued is to send the pickled result from #6 to the server and use it for further calculations

从 REST API 服务器和 examples/7Dimensional_optimization_and_plotting.ipynb 中的示例开始，重点关注 res.models[-1].sample_y(...) 调用。审查 #6 中描述的 pickle 结果方案，并确定应如何公开对训练模型执行的动态计算。当服务器能够执行所请求的模型计算而无需本地 Python 访问时，即表示完成。

由索引模型根据 Issue 内容生成。