Yes, the RS range of devices have this behavior on their MPPTs when working in limited current mode.
Generally speaking, the MPPTs doesn’t have the means to know, for each input, the capability of the strings, how many panels are connected and which is their Isc, in order to do a proper balance between strings.
For example, maybe one has string A capable of 3000Wp and string B capable of 2000Wp. And a consumption of 2500W. So a proper balance will be 1500W on string A and 1000W on string B.
If the algorithm will balance equally, as 1250W on both strings, then the string A will be “underused” and string B will be “overused”, even if they are in their working parameters.
More of that, there is the (partial) shadow problem, that will also complicate things.
So probably the developers have implemented a step by step methodology, ie., if a MPPT can supply a certain power up to a certain level, it will.
And only then will intervene the other(s) MPPT(s).
You will see that during time, at some point only MPPT2 will work and MPPT1 will not.
I don’t know if the choosing of the working MPPT is random or following some algorithm.
But in the end the system works and supply what is needed and this is the important thing.