Any comment in the sites Fire Risk Assessment or insurer correspondence?

Not from the brief info you have given. Outside, in cages away from the building - good. Presumably well ventilated? No ignition sources present? Impact protection from cars etc?

Calor will burn easily enough in air, so the oxygen won't make much difference. Highly unlikely both would leak at the same time.

The dispersion distance to below LEL will be very small for a pinhole leak as well, even allowing for oxygen being present.

Not sure what the leak frequency is for calor bottles, but if I recall correctly from when I did a DSEAR assessment for an aerosol filling company, they had failure rates of 1 in 100000. 

SO if you were to do an event tree to include leak probabilities, the result would be a very low probabilty.

Chance of leak x good ventilation x ignition source being present within LEL distance x both bottles leaking

I agree with Chris.  And even if the probability of simultaneous leaks is low what about a fire situation?  Or an incident in which both fuel and oxygen cylinders get damaged?  I can't look up the guidance myself because I have no general internet access at the moment (go figure its a govt internet system so I can look at bookmarked pages but nothing new - pain in the neck, I need to look at alternative FLT forks and grabs for handling fabric - guess I won't be doing any work today!!!).  The principle has to be to keep fuels and oxidisers as far apart as possible.