I can't imagine what is the basis for "efficiencies" on the order of 90% - 95% for a turbine. What kind of efficiency calculation are you using? I'm used to so-called enthalpy drop efficiencies for (non re-heat steam) turbines being on the order of 75% - 85%, depending upon the design details.
In any case, I'll list a few possible causes of higher-than-expected efficiencies, regardless of the magnitude of the efficiency numbers:
(1)As you recognized, infiltration leakage into the main flow can increase the apparent efficiency, but only if you are measuring flow (and section power)
(2)If the temperature measurements are in error, the efficiency error that results will be significant to the extent that the available energy across the IP section is small.
(2B) The temperature measurement error could be associated with the hypothetical leakage you'd mentioned: steam (of a different temp than the main flow) leaking from a seal area and onto or near your temperature instrumentation will make temperatures from the best-calibrated RTDs worthless.
(3) Is the IP turbine a reheat design? (This might explain the high efficiencies that you'd cited) Again the temperature measurements become critical; measuring the real main flow temperature, or a very well mixed flow in the piping leading into and out of the section.
In general, isolation (or proper accounting) of "side streams" in a large (i.e. 200MW) T-G set can be a difficult exercise. I assume that you have a working familiarity with ASME PTC6 (and 6A); if not these are "must reads" for you. (I continue to assume it's a steam turbine; don't know what else it might be. In any case, comments about errors still apply to any other working fluid.)
"Evaluating and Improving Steam Turbine Performance" - K.C. Cotton, 2nd ed. (Cotton Fact, Inc. ISBN 0-9639955-1-0, tel. 518.384.7885) is an excellent reference. Ken Cotton (deceased) worked for 30+ years at GE and was a key figure in the creation of the steam turbine performance test code, PTC6.
I hope that this is of some help.
