I don't think arguments about what language the rules would include support your side of the argument. We have, in the example of the War Magic feat and the history of Crawford's statements thereupon, clear and unequivocal proof that the same language and construction used in the Shield Master bonus action was intended to enable the use of a bonus action before or after the action upon which its use was conditioned. The fact that Crawford later changed his mind does not change the original intention that his published Sage Advice article revealed. At the time that these rules were written and the Player's Handbook was published, the language in question was not intended to impose a timing requirement.
I am puzzled and somewhat amused by the rigid mindset some folks seem to have about D&D. This whole issue of timing has never been a problem for me, even after getting into the semantic minutia on this topic. If the Attack Action happens "on your turn" you get to choose when to take the bonus shove "during your turn." The first thing just has to happen on your turn, while the second happens at a point of your choosing within that turn. It's as simple as it can be. Your turn is 6 seconds long, with a very limited range and number of individual activities you can do, so this cannot be very complicated.
You know that you can start with a shove, whether that attack comes from the Attack action or from a bonus action. The shove is exactly the same in resolution and effect regardless of how the attack is granted. Is it really so terribly mind-bending to determine which of the two formal game constructions, neither of which have an impact on the objective "reality" of the game's fictional world, granted that attack until after the dice are rolled? Is concurrent resolution of an action and a bonus action really such a heavy lift? Reading posts from Max, it sounds like every turn of every combat in his game is handled like an aircraft pre-flight safety checklist. I don't think that's the best way to play D&D, but hey... that's just, like, my opinion, man.
Also, I'll repeat what I said earlier about the Attack action. "With this action" doesn't support anyone's position in this argument. If you get lettuce and tomato "with this burger," it comes on the burger as part of it. If you get fries and a drink "with this burger," they are separate items. I think it is a safe bet that when the 5e PHB was written, the Wizards were not really prepared for the kind of hyper-literal, super-gamist analysis those rules recieve. It probably never occurred to them at the time that guys like Max would insist, apparently in earnest, that a healthy druid would be turned to dust by nit-picking the rule syntax, or that people would get confused by simply taking an action with an associated bonus action which together give your character a certain total number of attacks, and taking those attacks in whatever order the player wants.
I think the most puzzling and amusing element of this whole long discussion is the assertion that burdening the simple execution of this action-bonus action combo with extra verisimilitude-smashing timing requirements based on a particular interpretation of the syntax of the trigger for that bonus action, in the context of another particular interpretation of the syntax of the action that forms that trigger, all of which contravene the confessed original intent of the writers of the language in question, somehow adds simplicity and ease of use to the game. That's like adding another page to your tax forms in order to make it faster to prepare your return.
I suppose there are some DMs who value the comfort of having a rigid system, no matter how gamist, for procedurally adjudicating each turn in an inflexible step-by-step manner. Hey, if that's what makes you happy, dude, follow your bliss.