Fix accelerometer staleness and IMU list aliasing, free all servos on reset, GNU GPL v3 License

[SaintSampo]

Accelerometer returns stale values

Fixes the "accelerometer requests seem to return accumulated value" report. Reads looked correct in a loop but returned old data when called one-off.

Block Data Update is enabled, so the LSM6DSO's output registers don't refresh until both bytes of the previous sample have been read. The timer callback only drained the gyro, so the accelerometer registers stayed latched on a sample captured right after the previous read. Read once, turn the robot, read again, and you got the orientation from before you moved it.

The timer now calls _read_acc_gyro(), which burst-reads both sensors every tick, capping staleness at one timer period (~4.8 ms at 208 Hz). BDU stays enabled, so MSB and LSB still come from the same sample and burst reads can't tear.

IMU accessors returned their internal lists

get_acc_rates(), get_gyro_rates() and get_acc_gyro_rates() returned self.irq_v directly, so a caller who stored the result saw it mutate on the next call or asynchronously from the timer callback. They now return copies.

The copy would otherwise have landed in the 208 Hz timer path, so the register read moved into a private _read_acc_gyro() that updates irq_v in place and returns nothing. The timer allocates no more than it did before.

License

Adds GNU GPL v3.

[KalticCodes]

For publically facing functions, please use Sphinx-format doc commwnts for auto-generated API docs

[SaintSampo]

Done. I clarified and expanded IMU method docstrings in XRPLib/imu.py

Also removed the servo fix since frank did that in a different PR