inference outputs not featuring uncertainties

[nronce]

Hi LiteLoc team! Thank you for this excellent tool.

I noticed a discrepancy between the Nature Communications paper and the provided inference scripts. The paper states that the network predicts corresponding uncertainties for the localization parameters, but the .csv output is currently limited to 6 columns: frame, xnm, ynm, znm, photon, prob.

After tracing the code, I found that while xyzi_sig is correctly generated in LiteLoc.forward(), it is not passed to post_process() in loc_model.py, and therefore never reaches the saver_func in multi_process.py.

I have successfully extracted these values by:
- Updating loc_model.py to pass xyzi_sig to the post-processing step.
- Modifying liteloc.py to extract xyzi_sig at the detected p_index.
- Scaling the results in multi_process.py.

But it could be useful to make all the features broadly available without tweaking the code, in particular by using the csv writer function of help_utils.py in 'write localizations' mode (which does not work as it is downstream of the issue).

Many thanks!
N