> I think it needs to for analog.
It only has to be really-really close.
No mechanical contraption can guide floppy tape really-really-close at normal tape speeds.
Some very fast-wind tape decks do get air-lift. This is good for head wear, bad for sound, but we don't ask for perfect sound in fast-wind.
At the much higher surface speed of Winchesters, air and careful contour can give a stable ground-effect flying height that is really-really-Really close.
I don't believe there is any objection to very-high speeds: many things will just get better, but tape cost goes through the roof. On a disk, that's not a direct problem. However at 3600-7200RPM you only get a few milliseconds per rotation, compared to 5-20 seconds for Mellotron, EchoPlex, etc.
Many modern hard drives will NOT keep spinning if they can't read their own data. They spin-up, look for servo tracks to set speed and head position by, and if they don't see them (because you unplugged the heads from the drive brain) they cry "Error!" and shut-down. Older drives (think 100MEG) would self-spin without control info from the platter.
Most >400MB drives won't hold their head steady without servo signals on the platter. If you go WAY back to 20MB stepper-motor drives, they would hold their head steady naturally.
There is a direct ancestor, though I do not know the name. There was a mechanical spectrograph. It was a 12" platter with an iron band rim and a standard tape head riding on it with a light spring. There was also an 8" tall 4" diameter hub. You wrap damp paper around that. There was a stiff wire rubbing the paper, mounted on a lead-screw so it moved vertically up the 8". You recorded a short sound on the rim. It played back over and over. The lead screw was also coupled to a narrow-band filter tuning knob. The filter swept a range of frequencies while the lead-screw ran the wire up the damp paper rotating in sync with the recording. The output of the filter went to the wire and burned the damp paper in proportion to the filter output. It was trashed in the 1960s (which is how I met parts of one).
Assuming the drive will spin and hold its head, you only have little problems. The "oxide" on hard drives is blended for digital, without a smooth middle zone to give clean analog. And of course all analog magnetic recording sucks bad without a Bias Oscillator and careful biasing and EQ. Running super-high "tape" speeds eliminates one bit of EQ but you still need bias and integration. Also the heads will be tuned for hard digital rather than analog. Also the heads may be very low impedance at audio (their impedance rises with frequency, they are tuned to work at 5MHz-50MHz, so the impedance at 5KHz and below may be a fraction of an ohm.)
I know it is a cop-out, but you can really do most anything you have described with a $69-$500 digital delay box. And trust me: analog delay is a royal pain. Even the original nasty digital delays were a lot easier to use, and now that A/D D/A and RAM cost less than corn flakes, analog delay is for masochists. You "can", but you are not making music, you are playing with (and cursing at) machines.
The best uses for old hard drives:
One friend opened a full-high 5" drive and glued sandpaper to the platter, used it for minor sanding jobs.
There is a website where you just disconnect the wide traces to the "voice coil" on the arm and connect to a loudspeaker amplifier. Yes, the impedance is near right and it does "speak". Throw a 4 ohm 10 watt resistor in series just so you don't hurt the amp at high levels.
