RPI not reliable
I haven't checked, but I assume BT were advancing the argument that RPI is erroneous (which it is: the ONS announced they were dumping it as an official measure of inflation in 2013 due to its flawed use of an arithmetic rather than geometric mean).
The ONS did for a while publish an alternative index, RPIJ, which recalculated RPI using a geometric mean instead—but they have recently decided to abandon that on the basis that it was only designed to illustrate how flawed RPI is.
GIven that the government's official statistics authority very publicly states that RPI gets inflation wrong (an effect which compounds exponentially over time), it does seem to me quite right that pension liabilities be determined according to some other index that calculates it correctly instead (e.g. CPI). To do otherwise places a totally unaffordable burden on the fund.
Assuming that this point was litigated, then I guess the court determined the fund's contractual commitments cannot be varied to account for this (long term) error—despite the disastrous impact this will have on it and its members.