Aggregate SHAP importances from different models

A couple of questions on the SHAP approach to the estimation of feature importance.

I would like to use the random forest, logistic regression, SVM, and kNN to train four classification models on a dataset. Parameters in each training are chosen to give the best accuracy and precision for every model. A feature has a different magnitude of SHAP values in every model.

1. Are these differences meaningful, so as the feature indeed has different importance depending on an algorithm (RF vs. SVM vs. kNN...)?
2. Can I aggregate feature importances from these four models? For example, can I sum up SHAP values of every feature in all four models combined and have summary importance among all models?
3. If 2 is valid, do I need the same feature set for every model?

Please, also provide related references, if you can.

A related previous post, Is it valid to compare SHAP values across models?, unfortunately, got no answers.