[libc,kernel] Rewrite ptostr for speed, add thousands separator to printk

[ghaerr]

Rewrites ptostr precision timer routine to be much faster, using divmod routine. Same for kernel %k precision timing output: rewritten for speed to using divmod.

Adds %ld, %lu thousands grouping to kernel printk (like recently added libc printf).

Enhances some kernel output to use thousands grouping commas for large numbers.