Performance issue with getHeaderGroups() when using many virtualized columns (10k+) - TanStack