load_data should include a screening function to drop bad temperature data

[serbinsh]

I noticed that when running unprocData[to_convert] <- lapply(unprocData[to_convert], convert_temp) we dont either set any bad data (e.g. -30.27) to NA or drop (set to NA) any bad voltage data before the conversion.

We should set any bad temp data to NA at this stage in the processing

[DavidsonKen]

This seems to be similar to issue #10. I implemented this fictionality in sanity_check_fine, so bad data will be addressed. Not sure if we want to remove it in the load_data function as it seems more like a QAQC function to me