Before Harman, headphones like the HD600 and H650 were often used as benchmarks for comparing measurements. And they still are now. One potential problem with this is the effect of pad wear on the response of the higher frequencies (illustrated here). That could be mitigated though by using fresh pads for the calibration measurements.
There could be better options for this though.
I am basically just translating what Resolve and Cam said above. But for good calibration, what you probably need is a sound source that’s readily accessible, affordable, highly standardized, exhibits minimal unit and positional variation in its sound, that’s close to both the FR and acoustic impedance of the sound stimulus you intend to measure. For most people, this would probably eliminate free and diffuse field, since those require special gear and spaces for measurement.
If you’re using open headphones, then an open headphone that measures very consistently would probably be a good solution. If you’re measuring closed backs, then a closed back with good consistent sealing would probably be best.
Another option would be a pair of speakers with good directivity indices, EQ-ed to a specific in-room response, such as an approximately -0.85 dB/octave slope. This would require a dummy or real head and shoulders to interact with the sound field for the in-ear speaker measurements. And if you’re mostly measuring headphones, the acoustic Z might be a bit different.