[Enhancement] Automatically remove models not feasible to train when high-cardinality classification problem is presented

As a user, I would like AutoGluon to automatically remove models, which are not likely to be trained when high-cardinality classification is presented (i.e. 1M rows, 500 classes are likely make CatBoost model not feasible). User should be able to override defaults if the models are mentioned explicitly.

Start by locating AutoGluon's tabular classification model-selection entry point and the existing handling of explicitly requested models. Reproduce the high-cardinality case described in the issue, then define and verify behavior for automatically excluding infeasible models while preserving an explicit override; the payload names no files or tests.

Written by the indexing model from the issue text.