Many thanks for the excellent pictures and image enhancements, much beyond my skills.
The trajectory of the valve shows that the cylinder was still fastened in its original position when it departed -- caveat that a hull failure directly below may have sucked the bottle straight down and caused the valve to part company -- 5% probability?
The more likely case is that the valve departed while the bottle was still in situ. The equal and opposite reaction would have loaded the lower cup fitting which seems to be supported by two rods from the ends of the upper tank support. Note that both rods are gone and that a recent photo shows one rod for the adjacent tank hanging from its upper tank support but no longer connected to the lower cup. Note that the remaining rod was sufficient to retain that tank
It's been too many decades since uni physics for me to calculate the thrust from an 1800 psi tank with missing valve -- that would apply only if the tank held together.
A fragment comprising the bottom of the tank would likely have a higher velocity while an intact tank would have a greater duration of banging against the fuselage structure. But in the case of tank fragmentation, other bits of the tank would likely be found imbedded in baggage (caveat some sucked out) and adjacent structure, or at least marks, perhaps with green smears.
The hoop frames seem cleanly cut; so likely at a joint. Was that the bottle banging at them, or the baggage pushing them, or something dislodging a piece spanning the frames at the joints, or???
Note also that the metal curled up against the fuselage shows erosion of the interior paint in a number of panels. The boundary in one area shows the perimeter of a spray pattern -- oxidation of the paint???