Re: More timing info

[Moon@MIT-AI]

Here's some more accurate timing info for reading in LMIO;CHSNCP,
using the previously-mentioned optimization of the STRING-SEARCH-CHAR

The LINE-IN operations takes 24 ms on the average.
5 ms of this is spent in MAKE-ARRAY and ADJUST-ARRAY-SIZE (3.5 and 1.9 respectively)
10 ms, averaged, is spent getting a new packet.
9 ms of LINE-IN are not accounted for, this includes executing miscellanous Lisp
 code, searching for the CR, and copying the line once it is found.

Getting a new packet is done once every 12 lines, and costs 114 ms.
Of this, 33 ms are for updating the who-line (assuming it is only updated
 once, i.e. it doesn't have to go blocked.)
58 ms are spent in CHAOS:GET-NEXT-PKT (or 5 ms per LINE-IN)