Couple of questions or concerns here.

Why only 13 ML algorithms (compare models function) ? why not less and why not more ? What is the rationale behind this choice?

Also you are selecting an algorithm solely based on AUC and other accuracy metric. As far as I can see, it is too close to call !! You are deciding based on 3rd decimal point ! Also I don't think it is the right way to select an algorithm from an cohort of algorithms merely based on their accuracy metrics.

One also ought to see and understand how these ML algorithms function under the hood. What assumptions do they make. what statistical properties are these ML algorithms using.

Yes you might save time using low code solutions but eventually a misspecified model will cost you more both in terms of time and money.