Odd sampling frequency breaks the aligner

Using aligner.py and accelerometer.py to read data from sources with a sampling frequency that is not a multiple of 16 or if the numbers_to_samples is odd too, then after aligning some data may be left in the first key of self.data_map, which leads to batch_size being wrong. This breaks the aligner, since no subsequent package will have that batch_size.

https://github.com/INTO-CPS-Association/example-shm/blob/ce4724457b9c11f2513026d4009ca41b2bba489a/src/data/accel/hbk/accelerometer.py#L95-L104