The datasheet attached does not state any PWM frequency limit, thus the context as to why the OE recommends upto 1kHz is not known.
In theory you should be able to increase the switching frequency until some factor limits the ability to go any higher
- switching losses
- asymmetry in the RISE/FALL time of the drive signal path compounding the quality of the duty
- Higher switching frequency does not provide systematic benefit due to the inductance the load presents.
The switching losses might be important here and not just from the perspective of the drive FET but also from the solenoid because the solenoid core might be a solid iron core. This is valid for DC-fed as eddy currents are not really relevant and it means the core can be cast for economic sense. This however presents a problem once such a coil is energies by PWM as eddy currents will now be produced which will increase losses in the core but also present a MMF during the switching transient which can produce a significantly lower terminal inductance (sometimes as much as 1/100th) and thus the ripple current will be orders of magnitude higher.
One option is to include an output filter such that the coil sees a smoother DC being applied rather than a PWM voltage. This way the PWM can be increase to mitigate any audible concerns while also ensuring the coil is kept within the OE's recomendation.