Incorrect denominator in conversion of 16-bit signed values in readParmayMatrix #82
Description
When converting raw 16-bit signed integer readings from the accelerometer and gyroscope to g and °/s, the current implementation in readParmayMatrix() uses a fixed denominator of 32767 for all values. This does not account for the asymmetry of two's complement representation, which ranges from -32768 to +32767. As a result, negative values are slightly overestimated, introducing a systematic (very small) error.
Maximum error expected is around 0.2 m_g_ for acceleration and around 0.05 º/s for gyroscope, so it is negligible but it should be easy to fix.