I want to detect and delete duplicate files, but is there a way to be 100% that they are identical?
The ONLY way is to do a byte for byte check, and to also check for streams.
This can be done by first checking file sizes then comparing identically sized files, then checking for streams.
You CANNOT rely on hashes as they can collide.
Streams MUST be checked, because even if the main part of the files are identical, they could have streams attached, which are normally invisible to most software.
I do not know of any software that does this, but I'm sure there is some. Perhaps someone else can chime in there.