Fix not to modify bgmap by R0_thresh in bias step analysis

bgmap of BiasStepAnalysis is modified to -1 when estimated TES resistance R0 exceeds R0_thresh = 30e-3 (Ohm).
I suggest to remove self.bgmap[m] = -1.

The cause of this failure is not the cross talk but the wrong approximation when the TES is near saturation and the loop gain ~ 1.
And we can fix the estimation later using tes_param_correction.py.
We don't need to mask the channel in the bgmap.
Keeping the original bgmap used for bias step analysis is more reasonable and useful for future analysis.
https://github.com/simonsobs/sodetlib/blob/8f9593b5204cb64a777901f557fe2f6c5e38bb13/sodetlib/operations/bias_steps.py#L670-L685
https://github.com/simonsobs/sodetlib/blob/8f9593b5204cb64a777901f557fe2f6c5e38bb13/sodetlib/operations/bias_steps.py#L736-L742