IMHO, 'I don't want to go into details' means that you won't get any useful advice. The devil is in the details, and doling them out in dribs and drabs means that you are wasting people's time trying to guess at your situation.
The point about the busway being fed in the middle changes the picture, in a big way. What other little features might tip the scales in one direction or the other, hmm?
I do not believe that the NEC directly addresses your installation, but that you could make a case based upon guidance from the NEC that your installation is acceptable, or acceptable with small modification. Since the NEC is not directly applicable, you may be in a situation where you use it as the basis for an engineering analysis, and then use that engineering analysis to get an exception granted by whichever AHJ is involved, if any.
The first principle that you need to observe is 210.19(A)(1) 'ampacity not less than the maximum load served'. This is not an absolute rule; for some _specific_ non-continuous loads, the continuous ampacity of the conductor can be less than the short term loading created by the load device. 210.19 is about branch circuits, but it is a general principle followed all over the place.
The second principle is that if the conductors are sufficient for the load served, then in some specific limited cases, the OCPD may be sized larger than the ampacity of the conductors. Off the top of my head, the OCPD for motor wiring and welders are examples, and the reduced ampacity busway exception is another example. In general, the OCPD _must_ be sized to protect the conductors, but there are specific exceptions.
Clearly you couldn't safely put 900A of load on an 800A busway with 1000A of OCPD, regardless of the exception in 368.17(B). I don't see an explicit statement in 368 that says that a busway can only carry its rated current...presumably that is in the listing of the busway components
The exception in 368.17(B) clearly permits an installation where you might have an 800A busway with less than 800A of load on it, extending off of a 1000A busway, with the entire system is fed with 1000A OCPD. If additional devices are attached to this busway without proper supervision and load calculation, then there is a risk that the 800A busway would be overloaded without tripping the 1000A OCPD, and it is interesting to note that the 368.17(B) exception mentions industrial situations but does not mention 'engineering supervision'.
In your situation, if we imagined a slightly modified system, where a short length of 1000A busway was fed by the cable with the 1000A OCPD, and than two lengths of 800A busway were fed from the 1000A bus, the exception would clearly apply and your installation would be to code. Two 800A busways, each < 800A of load, 1000A OCPD shared by the two busways via a 1000A intermediate connection.
As I see it the problem is right at the connection between the cable and the busway. Here you have a connection that is presumably designed for the _800A_ busway, where you may have between 800A and 1000A of current. If you have 450A feeding one busway, and 475A feeding the other length of busway, then right at this junction you have 975A. I don't see how you can use the NEC to determine if this is safe. Presumably you would need some sort of engineering analysis to determine that this short length of busway can reliably and safely operate as an extremely short length of 1000A busway feeding the two 800A sections. This doesn't sound too far-fetched to me; while the actual bus bars are presumably 800A bars, you have considerable build-up of materials for actually making the junction. But this is no longer a question of code interpretation; it is a question of engineering analysis.
Please note: this is an opinion based on reading of the code and physics, not experience with these installations.
-Jon