Binary reading/writing has nothing to do with files that have a BOM. In fact, a BOM implies that the data is Unicode text, not binary data (unicode is not binary data). Realize that by supporting the BOM mIRC will be able to translate any UTF-16 into UTF-8 plaintext, so those files *used* to be seen as "binary" to mIRC, but not anymore thanks to this BOM support.
As for returning the matchtext, you should be able to use a subsequent call to $fread with $regex to match any text. I guess it could help to fill $regml when fseek -r is used, but I'm not sure how you'd return the matchtext of a wildcard.
Thirdly, $fread() isn't "forced" to read to the next line, this is simply the default behaviour, since it's the most common file operation. You can always utilize the switches to make it read N bytes, or read N to a bvar and set a %var to $bvar(&bvar,1-).text
