Exactly right, cfitz (as usual).
Packet written CD-R disks can be erased as you noted by removing their access although they remain on the disk.
There is no reason to suppose that multisession is more reliable than packet writing with a CD-R. In my opinion, the packet writing unreliability problem is essentially a CD-RW disk reliability problem, not the UDF file system used for packet writing (used on DVD's as well).
Multisession writing has its own potential reliability problems. See When should multisession recording be used?
Not even 99 sessions are available for multisession recording in the real world. The high overhead of this method uses up all available space before the theoretical 99 sessions can be used.
For instance, consider an 80 minute CDR:
The 702 MB capacity is for a 80 minute data disk written in the ISO 9660 format.
Packet writing on CD-RW disks is sometimes thought to have high overhead, but if you write more than seven sessions, packet writing is increasingly more efficient than multisession. On a 80 minute packet written CD-RW disk, about 571 MB of user space is always available regardless of the number of sessions. Using a CD-R disk, about 664 MB of user space is always available.
ISO multisession recording has very high overhead. Lead-in and lead-out for the first session uses about 22 MB and 13 MB for each subsequent session.
So if you want to write a lot of sessions, say 20, it would consume 269 MB of overhead and leave about 434 MB for user information (80 minute disk).
30 sessions would take 399 MB for overhead, leaving 304 MB for user information.
With multisession, the more sessions that you write, the less user space is available. Using an 80 minute CDR, you could write 52 sessions before you ran out of space. Unfortunately, the whole disk would be used for overhead, with no space available for user data.
Not too appealing, is it.