Re: ...accurate for a period of 158 million years...
From the top. The number of UT1 days is the number of transitions of IAU 1976 equinox across the Greenwich meridian (calculated by the proper motions of the stars in the FK5 catalogue using the IAU 1980 theory of nutation). The length of the UT1 day can vary from 86400 SI seconds by a few milliseconds so midnight doesn't happen on a whole second boundary.
By contrast, UTC days are designed to occupy an integral number of seconds, generally 86400, with extra seconds added or subtracted to keep UTC within 0.9s of UT1.
According to Wikipedia, the UK hasn't made up its mind whether GMT means UT1 or UTC so is there some ambiguity. But in all non-nautical contexts, I would expect GMT to mean UTC.
FWIW, a linear regression of the bulletin-B data from October 2009 suggests the next leap second will probably be end of June 2019. (It could just creep in at the end of December 2018.) But the Earth has actually speeded up in the 21st Century (we're had 19 fewer leap seconds than predicted in the late 20th Century) so we could end up having to add a few in a flurry, if the Earth regresses to mean. On the other hand, if the speed up accelerated, then we might end up having to drop seconds in the next couple of decades.
And if anybody thinks I'm wrong, please correct me because I'm working on our delta-T code right now!