I appreciate your thoughts and nod, nod and agree. you solved the pruning risk elegantly (keeping a copy of the block which contains my proof of existence tx), argument about keeping the document so why not bundle it with support data, it all makes perfect sense and I will adopt it in my argumentation as viable options of doing things.

as for keeping the document secret, that is ok 'cause my use case is about testaments and using rather higher fee but satoshi tokens for the stamping. now it can be reversed, using old coins, zero fees and spending them afterwards as usual.

how trezor fits here nicely is to sign message that describes the signing/timestamping with an address that would be used to actually timestamp a document. so that it all can be verfied in one eintopf of documents and signatures that are loosely related but can reveal much more info if the right off-blockchain data is added to it and is examined.