Hairpins doesn't work?

Those other changes are much simpler - 0they are one-time changes that don't have to happen in the middle of playback. Continuously varying volume in real time during playback is much, much harder. If it weren't, it would have been implemented long ago.

In fact it's rarely (if ever) true with acoustic instruments that a change in dynamic level simply produces a change in volume level alone. Of course, volume level does change when changing dynamic level, but acoustic instruments also change timbre (often quite significantly) at the same time. As I'm sure you know, most instruments sound much brighter (i.e. they include more high harmonic content) when playing loud than when playing quietly, and these harmonic levels usually change smoothly, gradually and continuously along with volume when changing dynamic levels over a held note. Not only that, but the attack characteristics of notes (especially those played by wind/brass instruments) can also change according to dynamic level.

Although Musescore uses velocity for fixed dynamic level markings such as pp, p, mf, f, ff etc., velocity alone is not sufficient for controlling playback of crescendo and diminuendo markings over a long note for the reason you mention (i.e. a velocity level is attached to the start of any given note and can not be changed during the course of that note - though it could be useful in controlling the attack envelope) BUT, neither is control of volume alone (because of the changes in timbre and attack mentioned above).

If it were to be implemented in Musescore, playback of crescendo and diminuendo marks over a long note obviously would require a continuous dynamic variation. I suggest that this would probably best be addressed with CC (continuous controller) messages, which can control more than one parameter at a time (e.g. volume + filter cutoff). In MIDI this is traditionally (though not necessarily) done with the "expression" parameter (CC11). Controlling volume alone is insufficient to produce anything approaching convincing playback of real instrument dynamics.

So, in Musescore, CC parameters could be used to help control the simulation of crescendo and diminuendo over long (held) notes by controlling volume and timbre simultaneously. However, in practice it's probably not that simple - using soundfonts, for example, those CC messages and velocity levels would need to be interpreted by the playback engine and changes in timbre/attack over held notes would need to be simulated either by (at the very least) changing the cutoff point of a filter in the playback engine smoothly (very important) and simultaneously along with the volume level on the same channel, or by using accurate multi-sampling with complex cross-fading and/or filtering as well. This would all need to be accurately controllable by the combination of velocity attached to the note plus the CC messages generated by your cresc./dim./ hairpin marks. I've no idea how much additional programming this would require in Musescore, but I suspect the developers currently have their hands full with bug-fixing issues more central to Musescore's stated priorites (i.e. creating high-quality readable notated music for human musicians to read).

I agree though, that it would be nice to have this feature at some point in the future...