OK. So I've done some more experimentation (and thrown away a bunch of discs) and have a little more empiric data, most of it not encouraging.
A LiveFS disc created and maintained by Windows IMAPI appears to be fundamentally different from one manually created using burning software such as Nero.
When Windows first "formats" a disc, and upon closing a session upon ejecting a partially written disc, it creates a new incomplete session which is used the next time the disc is written.
If Nero imports last written LifeFS session it works, but writing fails with a generic message that the disc structure is corrupted. I presume it is because Nero does expect/not know how to deal with the open session that follows the last closed session containing data.
Similarly Windows LFS does not know how to handle a mastered multisession disc (it treats it as blank), presumably because it is looking for an empty open session at the end, which Nero etc cannot do.
Windows LFS creates data sessions using UDF 2.01, and writes the discs in Mode 2 Form 1, which does contain some additional sector length for error correction, so it is not as bad as first thought. I can't figure how one could replicate that mode formatting on a mastered multisession disc using current versions of Nero.
I have found no other way to add files to a Windows IMAPI LFS disc other than Explorer/Drag+Drop.
Bottom line: I'n done playing with LFS.