[ink] use real units (1: time channel)

Sorry this comment is so late; I ran aground after rushing to get
earlier comments in by the deadline and this one was left in the
cognitive but not email-text buffer.

Reference:
http://www.w3.org/TR/2006/WD-InkML-20061023/

For example, in particular consider the time origin and units that are defined
for a time channel.

Time channel:
http://www.w3.org/TR/2006/WD-InkML-20061023/#time
Units:
http://www.w3.org/TR/2006/WD-InkML-20061023/#channel

So far as I can tell, although the example uses 'ms' which is an ISO
standard time unit designator, the only requirement that the InkML
format places on this units designation is that it be an xsd:string
value.  We should do better.

[coordinating approaches to representing timing is an interest of the
SYMM WG as I understand. There is precedent in SMIL, Timed Text, and
other places around W3C formats I imagine.]

Correlating the timestamps in the ink record with other timestamps or
time window requirements in other facets of the multimodal
interaction processing ensemble is too important to leave the time
units to the honor system like this.

Traces should use units that are taken from the ISO standard
namespace or be backed by metadata that gives the
highest-quality-available mapping between the times in the traces and
standards-based [ISO, UTC] timing data.

The format should instruct ink-trace-writers on how to document their
channel units, especially time, and indicate preferred units that
map directly to ISO and UTC, or to XSD, SMIL, or some other W3C
precedent.

Al