fix issue 103: use correct constant for gravity calculation

[amsjntz]

Implementing a fix for the gravity constant in the acceleration calibration process that caused the resulting z-axis offset to be incorrect. The gravity of 1G was being calculated like this: gravity = 32768 >> getFullScaleAccelRange();. getFullScaleAccelRange() returns 0 when the range of -2G to +2G is set, while the raw sensor readings go from -32768 to +32767. By that logic, 1G would equal about 16384.

Fixes #103

[Elizabeth-0]

I replicated the issue reported, and after running CalibrateAccel() with the current code, the Z-axis shows approximately 16364 instead of the expected value of ~8192 for the ±4G range.
This fix was previously mentioned in PR #94.

[amsjntz]

Hi, does the issue require further investigation, for example to check if other constants are off as well? The bug still exist in the currently available version of the library in the Arduino repository, which probably led to lots of headaches for many people trying to get their MPU6050 to work, so I imagine fixing it could be very beneficial