It’s a personal thing but I’d leave a bit more (-0.5 to -1dB). -0.1 is not leaving much “safety space”.
The results from converting will vary depending on what you use and the source & destination formats (which is why you need the margin). I’d suggest processing it, then doing a test conversion and looking at the results - load the converted file back in and see if the levels changed, and which way they went.
You can read up on maximizers via Google.