Development of libudf was based demand rather than any sort of attempt to implement all of the capabilities specified in the UDF standard or to define what might have been envisioned as a complete API.
It looks like you want to do is add a new facility that hasn't been previously needed by folks willing to work on this. This is not a slight on how useful or natural what you want to do is, just an artifact of how things worked out here.
A cursory look at the library seems to suggest that the file data is contiguous. So, from a UDF directory entry one can get a starting lba; based on a UDF blocksize and byte offset of where you want to get to, you could compute which sector you want to read from. In fact it looks like the read_sector routine in turn calls a byte-offset seek routine so you might be able to use that directly and bypass using the sector reading routine.
In sum, it looks like the building blocks for the API you want are probably there. But some coding and a new API function are needed if you want to do this more efficiently than what's currently provided.
libcdio has always been an open-source, community effort project. In fact, what's there has been put there by folks wanting something enough to work on it. So dig in!