This commit makes it such that the pooling allocator will be configured with a
much lower upper bound for memory pages, which will greatly reduce the
likelihood that the fuzzer memory limits will be hit from having too many
memories from too many instances committed.
da539255a5