We need to help first time users work out how to use our software, but I don't follow the logic on why we should prioritise around this. I get that we can lose users early on if they are confused by our apps, but that's not the full picture.
For a regular-use app (such as email in the example), what % of a user's time is spent as a new user, vs time spent as a no-longer-new user? Obviously over the lifetime of an app the amount of time spent as a new user is far less than that spent as a non-new user. After a few uses I know where the button is. But the design compromises (eg less space in the UI for content due to the oversize button) persist.
At some point the training wheels on the bike stop helping and start hindering.
This is the same gripe I have with the argument for UI animations "informing the user about what's happening". macOS (which stands out due to its refusal to just add a preference to fully disable animations) has educated me on the concept that an app minimises 'into the dock where it lives' many thousands of times now. I get it, honestly.
Maybe the solution is to have the UI grow in complexity as the user becomes more familiar? After the enlarged 'send' button has been clicked 5 times, reduce its size... maybe even do this gradually, a couple of pixels per click until it reaches 'expert size'. Or have an internal list of user actions and once a few of them have been completed offer to put the UI into intermediate mode?