Seems like a bit of an ideosyncratic use of the word. In tcmalloc these per thread zones are just called the thread cache (hence the name "thread caching malloc").
Glibc’s malloc has its own thread cache, which contains small chunks that can be accessed without locking at all: https://sourceware.org/glibc/wiki/MallocInternals#Thread_Loc.... (Interestingly, this cache used to be so optimized for speed at some point that it made a popular target for attackers, because it lacked many safety checks.)
