Support collations as defined in the new -15 caldav draft

[cdaboo@…]

The new -15 draft of caldav has added support for collations for the calendar-query report text-match operation. The server needs to support at a minimum the i;ascii-casemap and i;octet collations. i;octet is a straight octet-by-octet comparison (e.g. memcmp), i;ascii-casemap is a case-insensitive comparisons - however it must only casemap ascii characters < 0x80 - non-ascii must be left as-is. Unfortunately there is at present no collation that will do case-insensitive matching of non-ascii as well as proper unicode normalization. That has still to be defined. What's more I am not exactly sure what Python's string comparison operation currently does (that's what we have been using). That will need to be checked wrt unicode normalization.