Include params in .zip file

**Is your feature request related to a problem? Please describe.**
I may have a model that has been in use for a while. If I was not careful, the params are lost. When updating the model, and retraining, I may accidentally use different parameters or AutoML may arrive at different end result. This can lead to a model with different charasteristics. I think it especially happens when calibrator is used, the distribution on scores seems often very different across models.

**Describe the solution you'd like**
Option A) save params inside the zip. There is already a text file with the version number, so why not also training paramterers. One could also consider including performance metrics, training data stats, or even dataset name.

Option B) expose a way to load the model and then read the training parameters, at least those that exists, are readable or can be logically infered (such as how many leafs and trees)

**Describe alternatives you've considered**
It is possible to open the contents on the zip file in notepad, and sometimes I can see the params.

**Additional context**
Saving this information and especially dataset information may cause data leaks. For example a desktop app manufacturer might not want to make it easy to replicate his model. I think this is not big concern in my opinion, but including documentation on what is inside zip file and how to remove it can be useful. However, I do not think many users will extract zip files, perhaps even with .zip removed.