Change AM/PM to day periods

Deleted Component: xxx-spec

Bug 175 was to add structure to the data. However, in the course of doing so it
became apparent part of the bug should be delayed. This is for 2 reasons.

1. If we add aliases on the same level (as proposed in another bug) then a lot
of work goes away.
2. We can't do the calendars completely anyway because of the <am> / <pm>
fields. So this should wait resolution of another bug.

In particular if any calendar field is occupied, then you can't just alias the
whole thing. Eg.

1. Suppose we have a completely missing calendar. In that case, we can just
alias everything.

<calendar type="islamic-civil">
<alias source="locale" path="../calendar[@type='islamic']"/>
</calendar>

2. But if any subfield is present, we can just do that. We would like to have
<calendar type="chinese">
<alias source="locale" path="../calendar[@type='islamic']"/>
<dateFormats>...</dateFormats>
<timeFormats>...</timeFormats>
</calendar>

But that isn't legal. So what we would have to do is alias each subelement, eg

<calendar type="chinese">
<months>
<alias source="locale" path="../calendar[@type='gregorian']/months"/>
<months>
<days>
<alias source="locale" path="../calendar[@type='gregorian']/days"/>
<days>
<week>
<alias source="locale" path="../calendar[@type='gregorian']/week"/>
<week>
<eras>
<alias source="locale" path="../calendar[@type='gregorian']/eras"/>
<eras>
<dateFormats>
<alias source="locale" path="../calendar[@type='gregorian']/dateFormats"/>
<dateFormats>
<timeFormats>
<alias source="locale" path="../calendar[@type='gregorian']/timeFormats"/>
<timeFormats>
<dateTimeFormats>
<alias source="locale" path="../calendar[@type='gregorian']/dateTimeFormats"/>
<dateTimeFormats>
<fields>
<alias source="locale" path="../calendar[@type='gregorian']/fields"/>
<fields>
</calendar>
But even that doesn't cover everything since we can't alias the following:
<am>AM</am>
<pm>PM</pm>
(Which we've recognized is a bad structure: see 168 for possible fix to this;
then we could do the subelement by subelement approach if we wanted).

So this is being separated out from 175 for handling in 1.4.