[LinuxCNC/linuxcnc Issue#343] Apparent bug by inspection in mesa-hostmot2/encoder.c:hm2_encoder_read_control_register

The code is attempting to track the prior value of quad-error-enable. However, it does this by a single static value lasterrorenable. It looks like if not all encoders have the same quad-error-enable value, it won’t work properly, possibly leading to errors occurring and not being reported.

In particular, imagine a setup with 2 channels. channel 0 has quad-error-enable TRUE and channel 1 has quad-error-enable FALSE. Each time channel 0 is handled in the for-loop, lasterrorenable will be 0 because it was set the previous cycle when channel 1 was handled.