The team may well have a different answer but for me, the objectives of the HTML pages and the Preferences Pages aren’t the same, hence the reason for the different categories.
The HTML pages are a graphic representation / full list of all active Key Commands, including those that have been assigned by users. They’re a good way to learn active Key Commands and a quick way to find a Key Command you may have forgotten. The categorisation here is built around Modes and Functions - the exception is a Global Category that shows commands that are not assigned to a particular Mode or Function. Several of the Key Commands in this list are the same and that works because they deliver a different result depending on the Mode or Function.
Key Commands in Preferences are a list of (almost) all assigned and unassigned Key Commands. This is where users set or change Key Commands. With so many more (600+?) commands being listed, the categorisation here goes into more detail than the HTML Pages, breaking down modes into sub categories (e.g. Filter) and including oddities (UI, application etc,). It’s easier and more natural for users to work with this list of categories than it is to work with Key Commands being categorised in terms of their limitations.