if you found something that suggests only 1 unit of Pascal inaccuracy like you said, please let me know
"inaccuracy" is not the correct term to use.
The data sheet for the part states RMS noise of 1Pa in the high resolution filtered mode. You cannot detect pressure changes smaller than that. As suggested by this figure, I have verified that 8 cm differences in altitude are easily detected over the short term.
You should not confuse this spec with the spec
relative accuracy over pressure (range 800-1100 HPa) of 10 Pa, as that is irrelevant for an application like short term "hold altitude". The error in
barometrically determined altitude is limited entirely by atmospheric pressure variations, which are equivalent to +/- 50 meters or so.
So, if you want your drone to fly at 10 meters above ground, you have to zero the measurement, that is, determine the reference pressure at ground level, before every single flight.
