Return loss is the magnitude of the signal being reflected back from the connector, its phase is undefined.
You might expect two identical connectors to have the same phase of return loss, and so two -10dBs would add up to -4dB (*)(reflections are linear, not power, so add as an extra 6 dB, not 3) if the connectors were 'close' to each other. That may be true if the connectors have a poor return loss because of a consistent fault, like a design error, or being used in the wrong impedance system. If a good connector is nominally correct, then random manufacturing defects may cause it to have any phase of error.
Two identically bad connectors separated by a significant length of transmission line may, depending on the frequency (which controls the electrical length of the line) add in any phase, to get total return losses from -4dB to -infinity (*), or perfectly matched.
(*) I've neglected that the first connector, because of its return loss, transmits a little less signal to the second connector, so the two return losses will not add up to exactly -4 or exactly cancel, only approximately. As the return losses get better, the error in this approximation will drop dramatically.