Meoization and re-parsing in streaming

[FunMiles]

The original streamdown library documents their use of memoization and parsing caching to minimize the work done by the parser with streaming input.
Does svelte-streamdown do the same? The memoization of components is likely done automatically by svelte as it does not re-render components with unchanged properties. But how about parsing caching?
The docs say that svelte-streamdown has feature parity with the original. Does memoization and caching count as features? Can it be documented?