The uneven rule enforcement is by design. Twitter's report form says they only enforce their rules for 'protected categories.' The absurd result is what you see below.

Here's the language that appears when you report incitement or targeted harassment. By limiting enforcement to these categories, Twitter has rigged the system so that 'misgendering'—even in jest—is more egregious than explicit threats targeting sitting Supreme Court justices.

Twitter has thus equipped certain leftists with both a sword and a shield. They can target and threaten conservatives with impunity (views/values are not protected), while reporting you for hateful conduct (citing their identity) if you so much as joke about them in response.