referring to points 1 and 5: Please specify. “it doesn’t work” is not helping us to understand what might be the problem. Please post error messages, possible results and why you think they are not correct.
to point 2:
SNAP uses the quantifaction value from the metadata to convert the raw data to decimal values. This allows to store the raster data as integers (and saves lots of download time and disk space). You can read about it here QUANTIFICATION_VALUE of Sentinel-1C and here Radiometric conversion of sentinel 2 images
to point 3:
maybe the literature values refer to spectrometer measuremens (contact probe). Values from satellites depend on the spatial and radiometric resolution of the sensor, as well as on proper calibration. Have you checked if L2A data is available in your study area?
to point 4:
I am not familiar with the THOR viewer but SNAP offers the Spectrum View tool: Profile of a colour image Is this what you search for? If not, please specify.
About the library function: SNAP has no internal library, but you can use CSV data of spectral signatures in the Spectral Unmixing tool: Classification based on spectral library, also described here Weird result from the Spectral Unmixing of a Sentinel2 and discussed here Classification based on spectral library
to point 6: #
This is rather a question for the ENVI community:
ArcGIS is no remote sensing software. But there is a patch which promises Sentinel-2 support