The one in three is for overall force sizing - a long term average. It's not for figuring out how many ships will be available each week/month/year.
While not desirable, a short-term reduction may actually be less damaging than you'd think. There's an unusual confluence of programmes at the minute, which leaves more DD/FF in upkeep than is usual (T45 PIP and T23 SLEP/PGMU) compounded by a lack of industrial capacity (and funds) to execute those quickly. Losing a GP 23 or two saves you several tens of £M in upkeep funds, reduces the industrial burden so you get the more valuable ships (ASW T23) done earlier and the replacement date falls within the delivery window for T26/T31.
Not ideal, but some of those 23s really are in Triggers broom territory. Someone has (rightly) queried whether that money would be better spent addressing the future.